Device of image processing and method of image processing

FIELD: information technologies.

SUBSTANCE: image coding device determines a basic factor of quantisation (QPmb), serving as a predictable quantisation factor in case, when a generated code of the main coding approaches the target size of the code during coding of the input image. The image coding device codes the input image for each increment during control with feedback by performance of quantisation with the help of an adapted quantisation parameter (QPt) on the basis of an average parameter.

EFFECT: invention provides for compression of a generated code size for each image increment below the target size of a code in a proper manner without use of a quantisation factor serving as the basis for a quantisation pitch, with high deviation.

14 cl, 7 dwg

 

The technical field to which the invention relates.

The present invention relates, for example, to codereuse image device or the like, and more particularly relates to the technical field of the add-ons the size of the generated code to the target code size specified for a single image, without performing internal feedback control.

The level of technology

Still in the systems or the like for transmission of moving images or written to the storage media is highly efficient coding to efficiently provide the advantage of a transmission path or a recording capabilities. In embodying this encoding image device bit rate encoding of the bitstream produced by the encoder installed on some speed in accordance with the transmission rate of the transmitting medium and the size of generated data, that is, the quantization step for quantization in the encoder control with this limitation. That is, encodes image device compresses the amount of data that is generated due to the increase of the quantization step, for example, when continue image with a complex pattern, and Vice versa, when continue image with a simple pattern, fixed speed supported so as not to cause overflow or shortage, the internal filling of the buffer memory by reducing the quantization step and increase the size of the generated data.

Accordingly, in the image coding device according to this conventional method, when you have a complex image, the quantization step is increased, and when you have a simple image, the quantization step is reduced, and a uniform image quality in General unattainable. Considering this problem, for example, in Patent document 1 is disclosed encodes image device, where the selected amount of code assigned to each group of pictures, in accordance with the ratio of the sum of the complexity of coding for each group of pictures (GOP) and the complexity of coding for multiple groups of images is assigned to a greater extent for groups of images containing images with complex patterns of images, and is assigned to a lesser extent for groups of images containing images with simple patterns of images.

On the other hand, as a method of supplementing the size of generated code to the target code size specified for a single image, for example, a well-known stage 2 TM (test model 5). This method, in which the size of the code allocated for pictures, distributed by macroblock (MB) (MB), is taken as the target code size for the macroblock, and the feedback control used in the image.

In addition, when processing coding method, excessive compression, that the CSOs as MPEG or the like, encodes image device performs the processing of the quantization after performing an orthogonal transform such as discrete cosine transform (DCT) (DCT) or the like, allowing processing to reduce the amount of information. Then encodes the image device controls the value of this quantization for managing code size. At the same time supported a monotonically decreasing relationship between the quantization parameters and the size of generated code. Therefore encodes image device calculates the size of a code with a value of the quantization interval and puts a predicted value of the code size, located in an intermediate position, the linear interpolation or the like, whereby the size of the generated code, you can predict (binary search or the like to perform in the video, such as digital video or the like).

This method can be applied not only to the encoding method that uses a fixed table, such as MPEG2 or the like, but also to the way context adaptive coding used for AVC or the like.

Patent document 1: Japan patent No. 3358620.

However, when the above-described method, the step 2 DM when encoding the first picture of the sequence or picture, immediately after the scene change, the initial value of the quantization step is not consistent with the pattern of this picture, and, accordingly, there are cases when the image quality deteriorates.

For example, if the method in step 2 DM when encoding image device when the step of the quantization section until the appearance of the feedback is too large, the image quality of this part will deteriorate in comparison with other areas, and in the case when the magnitude of the quantization is too small, the size of the code used excessively in this area that may affect other areas.

In addition, in encoding the image device, the target code size for each MB is installed permanently on some code size, and is called an unacceptable distribution of the size of the code when there is an abnormality related to the difficulties of the image screen or the like.

Accordingly, the purpose of the present invention is to shrink the size of generated code for each increment of the image below the target amount of code in the right way without using the quantization factor used as the basis of the quantization step, with a large deviation.

The invention

In the imaging device according to the present invention for solving such problems is provided: block op is adelene basic quantization factor, made with the ability to determine the underlying factor of quantization, which is predicted so that the size of generated code for each increment of the image when encoding the input image will be approximated to the target code size for each increment of the image; a coding block, configured to encode the input image for each increment when the feedback control by performing quantization using quantization factor, defined on the basis of at least the basic quantization factor; and the power feedback control is performed with the opportunity to confirm the size of generated code for the input image coded by the coding block for each increment when the feedback control and in the case of predicting that the size of generated code for each increment of the image will be larger than the target code for each increment of the image, used to increase the quantization factor.

Thus, the imaging device used only increases the factor of quantization for limited conditions, thereby the fluctuation of the used quantization factor can be minimized.

In addition, in encoding the image method according to the present image the structure provided by the stages, to: determine a basic quantization factor to determine the basic quantization factor that is associated with the prediction that the size of generated code for each increment of the image when encoding the input image will be approximated to the target size of generated code for each increment of the image; encoding the input image for each increment when the feedback control to generate the coded stream by performing quantization using the used quantization factor, defined on the basis of at least the basic quantization factor; and control feedback to confirm the size of generated code for the input image coded by the coding block for each increment in the feedback control, and in the case of predicting that the size of generated code for each increment of the image will be larger than the target code for each increment of the image, used to increase the quantization factor.

Thus, encoding of the image method is used only increases the factor of quantization for limited conditions, thereby the fluctuation of the used quantization factor can be minimized.

According to the present invention used factor Quant is of only increases under limited conditions, due to which the fluctuation of the used quantization factor can be minimized. Thus, the imaging device and method of image processing can be realized, whereby the size of generated code for each increment of the image may be compressed below the target amount of code in the right way without using the quantization factor used as the basis of the quantization step, with a large deviation.

Brief description of drawings

1 is a structural diagram of the image coding device according to the first variant implementation of the present invention.

Figure 2 is a block diagram of the algorithm for a detailed description of the processing procedure coding respectively codereuse image device according to the first variant implementation of the present invention.

Figure 3 is a block diagram of the algorithm for further description of the processing of switching adaptive Q matrix.

Figure 4 is a conceptual diagram for describing what are corrected QP with discrete values, and the size codes are calculated by interpolation in respect of QP between them.

Figure 5 is a structural diagram of the image coding device according to the second variant of implementation of the present invention.

6 shows the I block diagram of the algorithm for describing the processing procedures concerning the processing of determining the base QP with feedback control.

Detailed description of the invention

The best embodiments of the present invention (hereinafter simply variants of implementation) will be described in detail with reference to the drawings. Note that description will be made in accordance with this sequence.

1. The first version of the implementation (thinking about the size of generated code using discretely selected quantization parameters)

2. The second variant of implementation (feedback control)

3. Other embodiments of the

1. The first option exercise

1-1. The features of the present inventions

Encodes the image device 100 and the method of encoding images according to a variant of implementation have such features, as described below.

Encodes the image device 100 applies the schema image compression using arithmetic coding, represented by the standard H.264/AVC (advanced video coding standard; advanced coding of moving images with compression). Encodes the image device 100 combines the parallel pre-encoding (predatirovaniya) and serial predatirovaniya. Therefore, encodes the image device 100 performs the predicting with great accuracy while reducing RA is a measure of the schemes and increasing lag (time delay).

Specifically, encodes the image device 100 performs quantization and part of the calculation code length along parallel first and second prekovremeno blocks 1 and 2, and other elements of the processing are used together, thus achieved a reduction schemes according to sharing schemes.

That is, the source, in the case of parallel pre-encoding, although it is necessary to provide all the processing elements in parallel, in encoding the image device 100 according to this variant implementation of the right is determined by the processing element, which can be used without affecting the accuracy of the model related to the processing element, is used. Thus the increase in size of the first and second predatious units 1 and 2 and the delay time (delay time) are suppressed.

In the first predatious block 1 parallel predatirovaniya, in which the accuracy is reduced, and the size of the circuits and the load processing is suppressed, runs on a wide range of parameters (QP) quantization, and the parameter QP quantization for implementation of the target code size is estimated approximately. In the second predatious unit 2 parallel predatirovaniya, from which the accuracy increases, runs narrow range with increased accuracy, and is determined by the base the parameter QP MBquantization used in the main encoding unit 3. As described above, encodes the image device 100 reduces the processing load, increasing the precision of the encoding of images. Further, the error due to simplification of the pre-encoding is the ratio of the bit rate and the parameter QP quantization, and, accordingly, a statistical model, as will be described in detail later, is created in advance, the error correction is performed based on the bit rate and parameter QP quantization.

Figure 1 illustrates and describes the configuration of the image coding device 100 according to a variant implementation of the present invention.

This encodes the image device 100 includes a first precederei block 1, the second precederei unit 2, the main encoding unit 3, unit 4 controls the size of the code and the buffers 5 and 6 of the delay.

First precederei block 1 is a module for performing the first pre-encoding, and includes a block 11 defining mode vnutriorgannogo prediction unit 12 processing vnutriorgannogo prediction unit 13 discrete cosine transform (DCT) (DCT), quantization block 14, block 15 calculating the length of the statistical code and the block 16 computing activity.

Second precederei block 2 is a module for performing a second predc is tiravanija and includes a block 21 processing vnutriorgannogo predictions the DCT block 22, block 23, the quantization unit 24 to calculate the length of a statistical code, the buffer 25, block 26 inverse DCT (ODCP) (IDCT) and the block 27 and the inverse quantization.

The main encoding unit 3 is a module for performing the main coding and includes the processing block 31 vnutriorgannogo prediction unit 32 DCT, block 33, the quantization unit 34 statistical encoding, the buffer 35, block 36 ADCP and block 37 of the inverse quantization.

Unit 4 controls the size of the code is a module for managing code size.

1-2. The prediction quantization parameter and the quantization matrices

1-2-1. The calculation of the size of generated code low precision

Encodes the image device 100 adaptively selects and uses, for example, three quantization matrix Q matrix in accordance with the complexity of coding (for a detailed description will be given later). Encodes the image device 100 sets a single quantization matrix Q matrix by simple processing of the first predatious block 1 parameters QP quantization in the range that can be accepted set by the quantization matrix Q matrix. Hereinafter, the size of generated code, evaluated as the results of the first predatious unit 1, will be called the size of generated code low fine the STI. Encodes the image device 100 performs the same processing for all quantization matrices Q matrix and calculates the size of the generated code low accuracy when changing the quantization matrix Q matrix and parameter QP quantization. In encoding the image device 100 parameter QP quantization and the quantization matrix Q matrix, in which the size of generated code low precision is the closest to the target code size, will be determined as the predicted parameter QPd quantization used as the average quantization parameter "Q" for pictures in the next step (second predatious block 2), and the quantization matrix Q matrix for the images used in the next step (hereinafter called the predicted quantization matrix Q D").

At the same time encodes the image device 100 uses the discrete part of the selected parameters QP quantization (hereinafter they will be referred to as the selected parameters QP1 quantization) to compute the size of the generated code low precision for the input image 91. Encodes the image device 100 calculates the size of the generated code low accuracy between the selected parameters QP1 quantization by interpolation, making it calculates the size of the generated code low precision is in respect of all the range of parameters QP quantization in the range which may be adopted by the quantization matrices Q matrix.

In practice, the input image 91 is first entered in block 11 defining mode vnutriorgannogo predictions in the first predatious block 1. Block 11 defining mode vnutriorgannogo prediction generates differential data of the image in all modes vnutriorgannogo predictions based on the input image 91, and determines the mode vnutriorgannogo predictions based on the predictions of the size of generated code data in the differential image. The predicted mode (predicted direction) is determined from the nine prediction modes in increments of a minimum of 4×4 pixels.

This specific mode vnutriorgannogo predictions is passed to block 12 processing vnutriorgannogo predictions, and is transmitted to the second precederei unit 2 and the main encoding unit 3. This mode vnutriorgannogo prediction is also used for the second pre-encoding the second predatious unit 2 for the main coding the main encoding unit 3.

Then, the processing block 12 vnutriorgannogo prediction calculates a differential image between the predicted image and the input image 91 to generate the data of the differential image. The predicted image used here is created from I the underwater image 91, to reduce processing. Thus, the first precederei unit 1 performs processing vnutriorgannogo predictions, due to which the block of the inverse quantization unit ADCP and the buffer can be reduced and the size of the circuits can be reduced.

When performing DCT processing with integer precision data on the differential image to generate DCT coefficients of the DCT block 13 passes them to the quantization unit 14. When performing quantization on the DCT coefficients to generate quantized coefficients to the quantization unit 14 transmits them to the block 15 calculate the length of a statistical code. Unit 15 calculates length statistical code applies the method adaptive to the context variable length coding (CAVLC) in respect of the quantized coefficients. According to this coding CAVLC can adaptively choose a highly efficient coding scheme in accordance with the surrounding conditions.

As described above, one of the signs of the first pre-encoding is that the coding CAVLC is used to calculate the size of the codes, even when the main encoding scheme statistical coding using adaptive to the context of binary arithmetic coding (SAVAS). Note that the encoding SAVAS is adaptive to the context of binary arithmetic encoding.

Here, the block 14 is Mantovani includes units 14-1, ..., 14-n (n=1, 2, 3, ...) quantization, which are connected in parallel, and the block 15 length calculations statistical code includes blocks 15-1, ..., 15-n (n=1, 2, 3,....) calculate the length of a statistical code, which is included in parallel. The value of n is set, for example, equal to 15. The quantization unit 14 sets in each block 14-1, ..., 14-n of the selected quantization parameters QP1 quantization corresponding to the set quantization matrix Q matrix of parameters QP quantization between 0 and 51. These selected parameters QP1 quantization discretely selected on a random interval in the range of parameters QP quantization, which can be accepted by the quantization matrix Q matrix. Note that the selected parameters QP1 quantization can be selected on some interval, for example, or you can choose the interval that varies with the value of the parameter QP quantization. With this configuration, the first precederei unit 1 performs quantization and calculation of the length of the code in parallel with respect to many parameters QP quantization, which is the same as described above for the parallel number, and gives each the size of generated code in the unit 4 control code size.

That is, the first precederei unit 1 performs the first predatirovaniya on a wide range of parameters QP quantization via computers is high, the pre-encoding, the consequence, reduce the size of the circuits associated with the quantization unit 14 and unit 15 calculates length statistical code, which helps calculate the size of generated code for a wide range of parameters QP quantization.

Here, the calculation block 16 activity calculates the activity in parallel with the determination of the mode vnutriorgannogo prediction unit 11 definition mode vnutriorgannogo predictions and groups each macroblock (MB) with this activity. That is, for example, assuming the case when grouped NumOfActivityGroup, block 16 calculation of activity compares the value of the activity thresholds ActivityThreshold[0] - ActivityThreshold[NumOfActivityGroup-2], thanks to which the determined group activity.

Note that the parameter QP quantization, actually used in the quantization process, is obtained by adding offset (AdaptQPDelta), which depends on the group's activity, to the middle parameter QP quantization (Q) for images.

MB_QP=Q+AdaptQpDelta[gruppens]

For example, if the number of group activity NumOfActivityGroup equal to 13, each value of the shift AdaptQpDelta can be defined as:

AdaptQpDelta[13]={-6, -5, -4, -3, -2, -1, 0, 1, 2, 3, 4, 5, 6}.

The number of group activity (ActivityGroup)defined for each MB, introduced in the quantization unit 14. Unit 14 calculates the quantization adaptive parameter QPt quantization Appendix the receiving shift according to the group activity to each of the selected parameters QP1 quantization. The quantization unit 14 performs the processing of the quantization on the DCT coefficients in the adaptive parameter QPt quantization.

First precederei unit 1 specifies the following quantization matrix Q matrix and the selected parameters QP1 quantization corresponding to the quantization matrix Q matrix, and thereby the size of the generated code is calculated in the same way. As a result of this first precederei unit 1 calculates the size of generated code for each of the quantization matrices Q matrix and the selected parameters QP1 quantization corresponding to the quantization matrices Q matrix.

1-2-2. Correction of errors in the size of generated codes

Unit 4 control code size adjusts the size of the generated code, calculated in the first predatious block 1.

Mentioned here the correction is performed taking into account the benefits that the error has some degree of trend.

The first cause of the error is due to the fact that the input image 91 is used to process vnutriorgannogo predictions instead of the locally decoded (Local Decode) screen. In this case there is no distortion due to the codec on the screen, used for vnutriorgannogo predictions, and, accordingly, there is a tendency in which the coding efficiency is superior coding efficiency when the real is the first encoding and generated a small amount of code. The size of this error depends on the size of the distortion, and, accordingly, there is a tendency in which the ratio between the size of the generated code and the error, the smaller the bit rate (Bit rate), the greater the error. In addition, when the ratio between the parameter QP quantization and error there is a tendency for the more parameter QP quantization, the greater the error. Accordingly, statistical data on size error should be prepared in advance and are determined as a function of bit rate r and the parameter QP quantization "q".

Specifically, the control unit 4 size code creates each of the average model error according to the bit rate and the average model error according to the parameter QP quantization in advance in relation to the size of the generated code, when the encoding is performed by CAVLC. This model error is stored in advance as, for example, a numerical expression or a table corresponding to each parameter QP quantization and bit rate. On the basis of the parameter QP quantization and bit rate block 4 control the size of the code calculates the correction values C_rate and C_qp respectively, representing the subject of error correction based on the corresponding model errors. From these values the correction C_rate and C_qp corresponding to the parameter QP quantization and bitoolshome, unit 4 controls the size of the code selects a smaller value as the correction values of Cv in accordance with the following equation.

Correction value Cv=min(C_rate, C_qp)

Thus, the value of error correction increases excessively, and, accordingly, adjusted the size of the generated code is prevented from increasing in comparison with the size of the generated code high precision, calculated in the second predatious unit 2. Note that the correction values C_rate and C_qp each point ratio (%) correction values for each of the size of the generated code (the code size of the pre-encoding), the calculated first predatious block 1.

Unit 4 controls the size of the code multiplies each of the size of the generated code, the calculated first predatious unit 1, the correction value Cv in accordance with the following equation, due to which calculates the amount of compensation in respect of the size of the generated code (hereinafter it is called the size-correction code CAVLC).

Code size correction CAVLC = code Size of the pre-encoding × Cv.

Unit 4 controls the size of the code adds the code size correction CAVLC to the size of the generated code, making it calculates the size of the generated code low accuracy in CAVLC.

The second reason for the error is that the error is produced which moves only when the means of statistical encoding choose CABAC. First precederei block 1 predicts the size of generated code low precision, when encoding is performed by SAVAS of code size CAVLC without performing encoding by SAVAS. SAVAS exceeds CAVLC respect to coding efficiency, and, accordingly, there is a tendency for the size of the generated code, precociously by CAVLC, the result is longer than the actual size of the code. For example, taking into account the ratio between the size of the generated code and the error, from a statistical point of view in the magnitude of this error there is a tendency for the smaller bit rate, the larger the size of the error due to improvements in efficiency SAVAS. Similarly, it is also adjusted by gathering statistics on the magnitude of the error in advance and create a model of the average error.

It is revealed that the error due SAVAS changes in the opposite direction with respect to the parameter QP quantization and bit rate, and the magnitude of this change is small compared to the error caused by coding according to the size of CAVLC code.

Therefore, the amount of correction with respect to SAVAS (hereinafter referred to as " correction SAVAS) is determined as a function of bit rate r and the parameter QP quantization "q". At the same time, the value of the Cb correction is calculated by the following equation.

Unit 4 controls the size of the code multiplies each of the size of the generated code (size codes pre-encoding), the calculated first predatious unit 1, the correction value Cv in accordance with the following equation, due to which calculates the amount of correction with respect to the size of the generated code (hereinafter referred to as " code size correction SAVAS).

Code size correction SAVAS = code Size of the pre-encoding × Cb.

Unit 4 controls the size of the code adds the code size correction SAVAS to the size of the generated code low precision, allowing to calculate the size of the generated code low accuracy in SAVAS. As shown in figure 2, unit 4 controls the size of the code can calculate each of the correction values SAVAS and size-correction code SAVAS (indicated by black circles) for each of the size of the generated code (indicated by squares), calculated first predatious block 1.

Then the control unit 4 size code performs the evaluation process for the quantization parameter (QP). As described above, the first precederei unit 1 obtains the size of the generated code by performing the pre-encoding using the selected parameter QP1 quantization with a discrete value on an arbitrary interval. Unit 4 control code size of vechicle the t size of the generated code by interpolation (indicated by white circles) in respect of the parameters QP quantization other, than the selected parameters QP1 quantization, in the range of parameters QP quantization, which can be accepted by the quantization matrix Q matrix. As a process of interpolation, you can use the usual process of interpolation, such as linear interpolation or the like.

That is, as shown in figure 2, the correction is performed in respect of the parameters QP quantization with discrete values obtained in the first predatious block 1 (indicated by squares)to obtain parameters QP quantization after the correction (indicated by black circles), and the code size is calculated by interpolation with respect to the parameters QP quantization between them (indicated by white circles).

Thus, the control unit 4 size code calculates the size of the generated code low accuracy in CAVLC by correcting the size of the generated code, the calculated first predatious unit 1, the error in the size of generated code generated in accordance with a simplified processing in the first predatious unit 1 relating to the size of the generated code first predatious unit 1. Thus the unit 4 control code size can improve the prediction accuracy of the size of the generated code by processing the simplified coding. By using the size of generated code in respect of CAVLC the Lok 4 control the size of the code uses the size of the generated code low accuracy in SAVAS to calculate the size of the generated code low accuracy, which is the predicted size for the size of the generated code due to SAVAS. Thus, the unit 4 control code size can estimate the size of the generated code low precision due SAVAS without performing SAVAS, because of which the processing is complicated. Unit 4 controls the size of the code predicts the size of generated code parameters QP quantization other than the selected parameters QP1 quantization, through a process of interpolation of the size of the generated code, the predicted selected parameters QP1 quantization, which is selected discrete. Thus, the control unit 4 size code no need to take the trouble of coding the input image 91 by using any parameters QP quantization, which makes it possible to simplify the configuration of the first predatious block 1.

1-2-3. Determination of the predicted quantization matrix

As described above, the size of generated code low precision is calculated for all parameters QP quantization, which can be accepted by the quantization matrix Q matrix. Unit 4 controls the size of the code replaces the quantization matrix Q matrix in accordance with the complexity of coding and selects based on the size of the generated code corresponding to the modified quantization matrix Q matrix, the parameter QP Kwan is Finance, which is the closest to the target code size, as the base parameter QPMBthe quantization.

Unit 4 controls the size of the code selects the parameter QP quantization used in the time when it turns out the size of generated code low precision, the closest to the target code size, as the nearest parameter QPn quantization for each quantization matrix Q matrix. Unit 4 controls the size of the code uses the closest parameter QPn quantization selected for each of the quantization matrices Q matrix, for example, as the complexity of the coding. It is clear that you can also use another indicator, such as the activity or the like. Here, the number of quantization matrices Q matrix to be switching to use, is accepted as NumOfMatrixId, where Id (identifiers) are assigned in descending order from a quantization matrix Q matrix with a slight gradient, and the maximum parameter QP in the range that can be accepted each quantization matrices Q matrix", is accepted as QMatrixThreshold[Id].

Unit 4 controls the size of the code compares the local parameter QPn quantization and QMatrixThreshold, starting from the quantization matrix Q matrix with a small value of Id. Unit 4 control code size determines the quantization matrix Q matrix with the smallest Id as predicted matrix quanta is of "Q D" among the quantization matrices Q matrix with neighboring parameter QPn quantization, less than QMatrixThreshold[Id]. Unit 4 control code size determines the closest parameter QPn quantization in the predicted quantization matrix Q matrix as predicted parameter QPd.

That is, the unit 4 control code size determines the quantization matrix Q matrix having the smallest gradient, as predicted quantization matrix Q D" among the quantization matrices Q matrix, which can take parameters QP quantization, in which the size of generated code low-precision approximate to the target code size. This predicted quantization matrix Q D" is also used for the main encoding the main encoding unit 3. Unit 4 controls the size of the code can use the quantization matrix Q matrix having the smallest gradient among those that meet the condition on the size of the generated code low precision, whereby deterioration in image quality can be prevented as much as possible.

Note that the unit 4 control code controls the first predatious unit 1 to calculate in order, from the quantization matrix Q matrix with small Id size of the generated code low accuracy in relation to selected parameters QP1 quantization in the range that can be accepted by the quantization matrix Q matrix. Then, in the case of the detection of the possible quantization matrix Q matrix with the closest parameter QPn quantization, less than QMatrixThreshold[Id], unit 4 controls the size of the code defines the quantization matrix Q matrix and the closest parameter QPn quantization as a predicted quantization matrix Q D" and predicted parameter QPd quantization. At the same time, the control unit 4 size code controls the first predatious unit 1 to start processing the next picture. That is, the control unit 4 size code controls the first predatious unit 1, in order not to calculate the size of the generated code low precision for quantization matrix Q matrix with the following Id in relation to the processed images. Thus, the unit 4 control code size can reduce the processing time required to determine the predicted parameter QPd quantization and the predicted quantization matrix Q D".

Thus, the control unit 4 size code selects from among the quantization matrices Q matrix, which can take parameters QP quantization, in which the size of generated code low-precision approaches to the target code size, the quantization matrix Q matrix with the lowest gradient as predicted quantization matrix Q D to prevent deterioration of image quality. Unit 4 controls the size of the code selects the parameter QP in the predicted quantization matrix quantum is s ' Q D", which the amount of generated code low accuracy becomes closest to the target code size, as predicted parameter QPd quantization. Thus the unit 4 control code size can be selected as the predicted parameter QPd of the quantization parameter QP quantization, which the amount of generated code low accuracy becomes closest to the target code size, when the adaptive change of quantization matrices Q matrix in such a way as to minimize the deterioration in image quality.

1-3. The definition of the base quantization parameter

Second precederei unit 2 actually performs the encoding is similar to encoder 3 by using the predicted parameter QPd quantization and the predicted quantization matrix Q D, and thereby calculates the size of generated code high accuracy with high precision. Hereinafter, the size of generated code, the calculated second predatious unit 2, will be called the size of generated code high precision. At the same time, the second precederei unit 2 calculates the size of the generated code high precision due to the use of not only the predicted parameter QPd quantization, but also parameters QP quantization prior to the predicted parameter QPd quantization followed, and predicts the size of the generated code high precision in from neighboring predicted parameter QPd quantization due to the use of speed fluctuations.

Unit 4 controls the size of the code gives the predicted parameter QPd quantization, the predicted quantization matrix Q D and group activity of each macroblock (MB) in the second precederei unit 2. Second precederei unit 2 performs a second predatirovaniya on the basis of these values.

In the second predatious unit 2, the input image 91 is introduced into the processing block 21 vnutriorgannogo predictions after passing the processing delay through the buffer 5 delay. The processing block 21 vnutriorgannogo prediction calculates the difference between the predicted image and the input image 91 to generate the data of the differential image. Then, the DCT block 22 performs the DCT processing on the data of the difference image to generate DCT coefficients. Unit 23 performs quantization processing of the quantization on the DCT coefficients to generate quantized coefficients. Unit 24 calculates length statistical code puts quantized coefficients statistical coding using CAVLC or SAVAS, making it calculates the size of the generated code high precision.

Note that in the processing in the second predatious block 2 block 23 quantization transmits the quantized coefficients in the block 27 and the inverse quantization. The block 27 and the inverse quantization applies the inverse quantum is their to quantized coefficients for playback of DCT coefficients. Then, the block 26 ADCP exposes the coefficients of the DCT transform ADCP, generating a locally decoded image, and stores it locally decoded picture in the buffer 25.

Here, the quantization block 23 includes three stage units 23-1, 23-2 and 23-3 quantization in this example. Unit 24 calculates length statistical code includes three stage units 24-1, 24-2 and 24-3 length calculations statistical code in this example. The reason why the number of stages is equal to three, is that the parameter QP quantization already roughly estimated first predatirovaniya on a wide range.

In this configuration, the quantization block 23 and block 24 length calculations statistical code run in parallel to get the size of generated code high precision with the predicted parameter QPd quantization and quantization parameters, the previous predicted parameter QPd quantization and the next. At the same time, the block 24 length calculations statistical code chooses the same method as the method of statistical coding for native encoding, the main encoding unit 3, from any of the CAVLC or SAVAS.

Following this, the control unit 4 size of the code defines the base parameter QPMBquantization for images that are used in the main encoding, the size of the generating system is imago code high precision, received at the second predatirovaniya. Then the control unit 4 size code transmits the information on the quantization (such as "Q matrix and QP for each MB from this particular basic parameter QPMBquantization for images, the predicted quantization matrix Q D and group activity each MB in the main encoding unit 3.

More specifically, in the case when the target code size is between the size of the generated code high precision obtained by the second predatirovaniya, that is, if Genericogeneric (QP_1+1)≤Aleviate ≤

≤Generated bits (QP_1-1),

the parameter QP quantization, which is the closest to the target code size is selected as the basic parameter QPMBthe quantization.

Otherwise, the control unit 4 size code obtains the rate of change of the size of generated code high precision with respect to changes in parameter QP quantization of the results of the second pre-encoding. The predicted parameter QPd quantization is calculated based on the size of the generated code to the low accuracy obtained first predatious unit 1. Therefore, the parameter QP quantization, the closest to the target code size, exist side by side with the predicted parameter QPd quantization. In the case when the parameter QP quantization close poznaniu, the rate of change in the amount of generated code is almost constant. Therefore, the control unit 4 size code predicts the size of generated code high precision for each parameter QP quantization of the rate of change in the size of generated code high precision with the predicted parameter QPd quantization and for parameters QP quantization prior to the predicted parameter QPd quantization followed, and selects the parameter QP quantization that is closest to the target size, as a basic parameter QPMBthe quantization.

First, when the parameter QP quantization subtracts "1", the value of Razatos, showing the percentage changes the size of generated code high precision, is, as follows from the results of the second pre-encoding 2. Note that the value Genericsite shows the size of the generated code in the second predatirovaniya 2, QP_1 shows the predicted parameter QPd quantization value QP_1-1 shows the parameter QP quantization, which is less than the predicted parameter QPd quantization, to "1".

Resnames=(Genericsite(QP_1-1)-Generify(QP_1))/(Genericsite(QP_1)).

When the parameter QP quantization is added to "1", the value of Razatos, showing the percentage change of the size generated by the th code high precision, is, as follows from the results of the second pre-encoding 2. Value QP_1+1 shows the parameter QP quantization, which is more than the predicted parameter QPd quantization, to "1".

Resnames=(Genericsite(QP_1)-Generify(QP_1+1))/(Genericsite(QP_1+1)).

The rate of change of Razatos in the size of generated code high accuracy close to the predicted parameter QPd quantization, are as follows.

Resnames=(Razatos+Resnames)/2.

That is, Razatos is calculated as the average of the changes in the size of generated code when the parameter QP quantization is changed to "1" in each of the positive and negative directions from the predicted parameter Q quantization.

Believing that QP is accepted as the absolute value of the difference between the parameter QP quantization, in which the size of the generated code high precision becomes closest to the target code size, and the predicted parameter QPd (QP_1) quantization, when the parameter QP quantization corresponding to the size of the generated code high precision closest to the target code size is less than the parameter QP (QP_1-1) quantization, which is less than the predicted parameter QPd quantization to 1, the size of generated code high precision (Genericsite(QP)) for the parameter QP quantized what I when the size of generated code high precision becomes closest to the target code size is calculated as follows.

Genericsite(QP)=Genericsite(QP_1-1)×(1,0+Resnames)^(QP-1)

When the parameter QP quantization corresponding to the size of the generated code high precision closest to the target code size is greater than the parameter QP quantization (QP_1+1)that is greater than the predicted parameter QPd quantization to 1, the size of generated code high precision (Genericsite(QP)) for the parameter QP quantization, in which the size of the generated code high precision becomes closest to the target code size is calculated as follows.

Genericsite(QP)=Genericsite(QP_1+1)×(1,0-Resnames)^(QP-1)

That is, the control unit 4 size of the code increases or decreases the size of the generated code high precision during use parameters QP quantization prior to the predicted parameter QPd quantization followed, the code size according to the rate of change when the value of the parameter QP quantization is changed to "1" relative to the predicted parameter QPd quantization as a mid. Unit 4 controls the size of the code can calculate with high accuracy the size of generated code high accuracy when used parameters QP kVA is tovani, closest to the predicted parameter QPd quantization.

As described above, the control unit 4 size code selects the parameter QP quantization, the closest to the target code size, as the base parameter QPMBquantization for use as the average quantization parameter (QP) when the main encoding.

As described above, the second precederei unit 2 performs predatirovaniya with the predicted parameter QPd quantization (QP_1), estimated first predatious block 1, parameter QP quantization (QP_1+1), a large 1, and the parameter QP quantization (QP_1-1), reduced by 1. Here, to reduce the size of circuits, as described above, given in parallel only the quantization block 23 and block 24 length calculations, statistical code, and other processing is used.

At the same time, the local decoding image used when processing vnutriorgannogo predictions, represents the data quantized with the predicted parameter QPd quantization (QP_1), estimated on the basis of the results of the first predatious block 1. That is, data to be processed in a reverse quantization and ADCP represent the quantized output of the predicted parameter QPd quantization (QP_1). This means that the inputs for processing vnutriorgannogo predictions when predecode the implement with parameters QP quantization (Orpedo+1) and (QP_1-1), prior to the predicted parameter QPd quantization and the next, are not locally coded images of these quantization parameters, but are replaced by a locally decoded image of the predicted parameter QPd quantization (QP_1).

Thus, by predicting the size of the generated code low accuracy on the basis of the results in the first predatious block 1, block 4 control the size of the code calculates the size of the generated code high precision by the same encoding as the primary coding according to the predicted parameter QPd quantization, for which there is an extremely high possibility that the size of generated code high precision becomes closest to the target code size, and with the previous and subsequent parameters QP quantization. Thus the unit 4 control code size can almost accurately calculate the size of the generated code high precision when using the predicted parameter QPd quantization and its preceding and subsequent parameters QP quantization. Further, with the present invention it is noted that the rate of change of the size of generated code high precision, accompanying the change of the parameter QP quantization, is almost constant in a narrow range. Unit 4 controls the size of the code calculates the size of GE is erasemode code high precision, when used parameters QP quantization closest to the predicted parameter QPd quantization, based on the predicted parameter QPd quantization and rate of change of the size of generated code high precision with its preceding and subsequent parameters QP quantization. Thus, the unit 4 control code size can almost accurately calculate the size of the generated code high precision for parameters QP quantization adjacent to the predicted parameter QPd quantization.

1-4. The main coding

The main encoding unit 3 performs the main predatirovaniya through the use of basic parameter QPMBquantization, based on the results of the second predatious unit 2, and the predicted quantization matrix Q matrix, obtained on the basis of the results of the first predatious block 1, group activity, mode vnutriorgannogo predictions and the like. Specifically, in the main encoding unit 3 when receiving the input image 91 which has been subjected to the processing delay through the buffer 6 delays in processing block 31 vnutriorgannogo prediction data of the differential image between the predicted image and the input image 91 is calculated mode vnutriorgannogo predictions found during the first predc is tiravanija. Unit 32 performs DCT the DCT process and the block 33 performs quantization quantization of DCT coefficients. Output (quantized coefficients) unit 33 quantization is transmitted in block 37 of the inverse quantization. Block 37 inverse quantization applies inverse quantization to the quantized coefficients to reproduce the DCT coefficients. Then the block 36 ADCP exposes the coefficients of the DCT transform ODCP to generate the predicted image and saves it predicted image in the buffer 35.

Thus, the statistical coding is performed by block 34 statistical coding after passing through DCT unit 32 DCT and quantization of DCT coefficients by block 33 DCT as described above, and displays the output stream 92 mounted on the target code size.

1-5. Conclusion

Therefore, encodes the image device 100 according to the present variant implementation during the first predatirovaniya first predatious unit 1 approximately calculates the size of the generated code low accuracy when the coding is performed with many of the selected parameters QP1 quantization, due to what is estimated predicted parameter QPd quantization serving as the target code size.

At the same time, the first precederei unit 1 performs parallel encoding, Thu is to calculate the size of the generated code in case when the encoding is performed with the selected parameters QP1 quantization on an arbitrary interval.

Usually, to perform encoding in parallel, the size of the circuits becomes very large. However, this variant implementation solves this problem by performing processes more generally when implementing parallel coding to reduce the size of the schema.

That is, more specifically, in the first predatious unit 1, as described previously, only the quantization unit 14 and unit 15 calculates length statistical code provides running in parallel, and other processing are implemented together. In addition to the exception block of the inverse quantization unit ADCP and buffer handling vnutriorgannogo prediction is performed using the input image as a predicted image.

On the other hand, when the statistical coding is used SAVAS, processing is not always performed with high bit rate. That is, in the case of experimental use SAVAS as statistical coding, when you calculate the size of the code with the given parameter QP quantization for predicting the size of the code, if the code size parameter QP quantization more (i.e. if the value of each quantized coefficient is greater), it is impossible to forecast the TB size of the generated code.

SAVAS is a method of statistical coding, in which the calculation of the probability for each single bit, thanks to which data is compressed. The processing for a single bit means that as the code size becomes larger, the processing time becomes longer, and accordingly, the processing is not completed within a certain period of time (e.g., one frame). Therefore, in the case of experimental use SAVAS as statistical coding, and when the code size is large, the first precederei block 1 difficulty in calculating the size of the generated code after statistical coding and difficulty in predicting the size of generated code at a high bit rate.

From this point of view in the first embodiment, the first precederei unit 1 uses CAVLC to calculate the size of the code. Thus, in the first predatious unit 1, the size of the circuits is reduced, the processing is reduced and used simplified predatirovaniya, allowing approximately estimated size of the generated code low precision. Then the control unit 4 size code selects the parameter QP quantization with the size of the generated code accuracy is low, close to the target code size, as predicted parameter QPd quantization.

In the second predatious unit 2 taking into account the fact that the assessment of the predicted parameter QPd quantization first predatious unit 1 includes an error, predatirovaniya is performed again, thereby improving the prediction parameter QP quantization. Specifically, predatirovaniya is performed again with parameters QP quantization in the vicinity of the predicted parameter QPd quantization, approximately evaluated first predatious unit 1, thanks to get the size of generated code high accuracy and parameter QP quantization, the closest to the target amount of generated code high precision, again serving as the target. When calculating the length of the statistical code is used the same way (SAVAS or CAVLC), and that the main encoding.

Note that although there is an error due to the difference in the input (locally decoded) image processing vnutriorgannogo predictions, parameter QP quantization has almost the same value, and the distortion due to encoding is also practically the same and, therefore, negligible.

1-6. Procedure

Below with reference to the flowchart of the algorithm in figure 3 will be described procedure RT1 encoding the encoding of the image device 100 according to a variant implementation of the present invention. Some or all of this procedure is also equivalent to the way the encoding images according to a variant implementation.

First block 16 calculation activity calculates the activity of individual MB shared the MB group activity in accordance with their values (step S1).

After this, the block 11 defining mode vnutriorgannogo prediction determines the mode vnutriorgannogo predictions based on the input image 91 (step S2). This mode vnutriorgannogo prediction is also used for the second pre-encoding the second predatious unit 2 for the main coding the main encoding unit 3.

Then, the processing block 12 vnutriorgannogo prediction calculates data of the differential image between the predicted image and the input image. Here as the predicted image to reduce the processing uses the input image. Next, block 13 performs DCT DCT integer precision and sends the DCT coefficients in the quantization unit 14 (step S3).

The quantization unit 14 quantum values of DCT coefficients of the set of selected parameters QP1 quantization on an arbitrary interval as the average quantization parameter (QP) for pictures. Block 15 calculate the statistical code length encodes the quantized coefficients into variable length codes and computes the length of a code, allowing it turns out the size of generated code for the selected option is in QP1 quantization (step S4). At the same time, as described above, the parameter QP quantization for each MB gets a value that takes into account the activity, and encoded. That is, as described above, the parameter QP quantization for each MB is obtained by adding offset-dependent group activity, to the average quantization parameter (QP) for images.

Note that in the case of processing of switching an adaptive quantization matrix Q matrix, the above processing is performed for each quantization matrix Q matrix. Specifically, predatirovaniya is performed with the selected parameter QP1 quantization discrete (discontinuous) value to obtain one of the art advantages of the size of generated code for each quantization matrix Q matrix. At the same time, the selected parameters QP1 quantization chosen to cover the range of parameters QP quantization, which can be adopted for each quantization matrix Q matrix.

Then the control unit 4 size code performs the correction processing on the size of the generated code, the calculated first predatious unit 1, due to which calculates the size of the generated code low precision. Unit 4 controls the size of the code corrects errors due to simplification of pre-encoding, and calculates through a process of interpolation size generiruemogo the code low accuracy, the corresponding parameters QP quantization other than the selected parameters QP1 quantization (step S5).

Unit 4 controls the size of the code performs the processing at step S5 in respect of each quantization matrix Q matrix to compute the size of the generated code low accuracy for each quantization matrix Q matrix (step S6). When the above processing get the size of generated code low accuracy that meets all the necessary parameters QP quantization, and, respectively, of the parameters QP quantization, which can cause the size of generated code low precision, the closest to the target code size, choose the quantization matrix Q matrix with the lowest gradient as predicted quantization matrix Q D". Next, the control unit 4 size code selects the parameter QP quantization, which can cause the size of generated code low precision, the closest to the target code size corresponding to the predicted quantization matrix Q D", as predicted parameter QPd quantization (step S7). In addition, the choice of the quantization matrix Q matrix, as mentioned above, limits the range of parameters QP quantization, which can be taken, and, accordingly, it is possible to reduce the range of selected parameters QP1 quantization for computing the size of a generated by the ode to a low accuracy in the first predatious block 1. They serve as the predicted quantization matrix Q D" and predicted parameter QPd quantization, which are defined in the first predatious block 1.

After this second precederei unit 2 performs processing of obtaining the size of the generated code (steps S8-S10). The purpose of the second predatious unit 2 is to improve the accuracy of the estimate of the underlying parameter QPMBquantization by performing the pre-encoding again, due to the fact that the evaluation of the predicted parameter QPd quantization first predatious unit 1 includes an error.

Specifically, predatirovaniya is performed again with parameters QP quantization in the vicinity of the predicted parameter QPd quantization, approximately estimated in the first predatious unit 1 and thereby gain the size of generated code high precision, and the parameter QP quantization that is closest to the target size is again. Calculating the length of the statistical code uses the same method (SAVAS or CAVLC), and that the main encoding.

Specifically, the control unit 4 size code performs the processing vnutriorgannogo prediction unit 21 processing vnutriorgannogo prediction and DCT unit 22 DCT mode vnutriorgannogo predictions found in the work of the first predatious b the eye 1 (step S8). Second precederei block 2 shares locally decoded image with the predicted parameter QPd quantization (QP_1), estimated in the first predatious unit 1, as a locally decoded image (predicted image), to be used in internal prediction.

When quantization is used predicted parameter QPd quantization (QP_1), the quantization matrix Q matrix and group activity found in the work of the first predatious block 1. The predicted parameter QPd quantization (QP_1) is installed in the unit 23-1 quantization, the predicted parameter QPd quantization (QP_1-1), on which "1" is smaller than the predicted parameter QPd quantization, is installed in the unit 23-2 quantization, and the predicted parameter QPd quantization (QP_1+1), on which "1"is greater than the predicted parameter QPd quantization, is installed in the unit 23-3 quantization.

Further, the parameter QP quantization for each MB gets a value that takes into account the activity, and encoded. According to the following second predatirovaniya you can get one from the picture of the advantages of the size of the generated code (step S9).

After that, the control unit 4 size code finds the basic parameter QPMBquantization of the size of the gene is dummy code high precision, based on the work of the second predatious unit 2 (step S10).

Following this, the main encoding unit 3 performs basic coding (step S11). When the main encoding the actual coding is performed using the basic parameter QPMBquantization for images found in the work of the second predatious unit 2, and the predicted quantization matrix Q matrix and group activity found in the work of the first predatious block 1. Thus the number of treatments associated with the encoding ends.

Below with reference to the flowchart of the algorithm in figure 4 will be given a further description of the procedure RT2 determine the Q matrix is performed in the step S7, the procedure RT1 of the encoding device 100, the image encoding.

When this processing, the control unit 4 size code first initializes the Id on Id=0 (this S21), and then compares the parameter QP quantization, in which the amount of generated code low precision is the closest to the target code size of the quantization matrix Q matrix with the lowest Id value, with a maximum parameter QP quantization (QMatrixThreshold[Id]), which can be taken in the quantization matrix Q matrix (step S22). Further, in the case when the parameter QP quantization, in which the amount of generated code low precision is the closest to the target issue for lighting the th code in the Id-th quantization matrix Q matrix, less than QMatrixThreshold[Id], block 4 control code size determines the current quantization matrix Q matrix as a predicted quantization matrix Q D". Next, after determining the parameter QP quantization, in which the amount of generated code low precision is the closest to the target code size in the predicted quantization matrix Q D", as predicted parameter QPd matrix quantization unit 4 controls the amount of code ends the procedure RT2 of the definition of "Q matrix.

On the other hand, at the step S22 in the case when the parameter QP quantization, in which the amount of generated code low precision is the closest to the target code size in the Id-th quantization matrix Q matrix, equal to or greater than QMatrixThreshold[Id], unit 4 controls the size of the code gives the increment Id (step S24). Unit 4 controls the size of the code defines smaller Id than the total number of quantization matrices Q matrix", "1" (NumOfQMatrixId-1) (step S25). Next if Id=NumOfQMatrixId-1 is not available, the unit 4 control code size is returned to step S22, and checks the next quantization matrix Q matrix. On the other hand, in case when Id=NumOfQMatrixId-1, block 4 control code selects the quantization matrix Q matrix with the steep gradient (quantization matrix Q matrix whose Id is equal NumOfQMatrixId) (step S23), and the procedure RT determine the Q matrix ends.

According to the procedure RT2 of the definition of "Q matrix in figure 4 block 4 control code size sets the maximum available parameter QP quantization for each quantization matrix Q matrix and determines the order from the quantization matrix Q matrix with smooth gradient, whether or not the size of generated code low-precision corresponding to the parameter QP quantization, in which the amount of generated code low precision is estimated as the closest to the target code size, specifies the value that is closest to the target code size. Then, in the case of specifying a value closest to the target code size, block 4 control code size determines the corresponding quantization matrix Q matrix as predicted quantization matrix Q D"to be used when the main encoding.

As described above, in the embodiment of the present invention are two encoding images in increments. To improve the efficiency of coding of the image device 100 increases the amount of processing and to resolve the problems, which consequently increases the size of schemes for encoding applies partially parallel configuration according to partial community schemes. Thus, while simplifying the configuration of precoders, correct accompanied by this simplification error for the odd statistical data.

Accordingly, encodes the image device 100 may add the amount of generated code of the primary encoding, which occurs when the main encoding target code size specified for a single image, without performing internal feedback control. Thus encodes the image device 100 can prevent problems with feedback control, such as adverse effects due to inappropriate parameter values feedback and improper distribution of the target code size, or the like. As a result of this encodes the image device 100 can determine the size of generated code of the primary encoding, in which the amount of generated code of the primary encoding is consistent with the target code size and takes into account the visual characteristics, i.e. a suitable quantization parameter.

Note that the present invention is not limited to the above-described embodiment, and various improvements or modifications without departing from the essence of the present invention.

For example, the above encodes the image device 100 and a method of coding images can also be implemented as a computer program installed in the device, the storage medium on which is recorded as the computer the NSS program, realizing the way, or a storage medium on which is recorded the program.

1-7. The operation and advantage

According to the above configuration encodes the image device 100 serving as the imaging device, encodes the input image 91 quantization of at least the input image 91 as the input image on the basis of DCT coefficients representing just coded data obtained by coding the input image 91 by simple processing, and based on the selected parameters QP1 quantization, discrete selected parameter QP quantization, which are the quantization factors, and calculates the size of the generated code of the encoded input image 91.

Encodes the image device 100 corrects the error in the amount of generated code of the input image upon coding, which is in accordance with simple processing, and calculates the size of the generated code low precision.

Encodes the image device 100 calculates the size of the generated code low accuracy at the same time as the input image 91 is encoded based on the parameters QP quantization other than the selected parameters QP1 quantization, through a process of interpolation in terms of the size of the generated code accuracy is low when the input is the images 91 is encoded based on the selected parameters QP1 quantization.

Therefore, encodes the image device 100 can calculate the size of generated code low accuracy on the basis of all parameters QP quantization without coding all parameters QP quantization, so that is possible to simplify the calculation of the size of generated code low accuracy and reduce computational scheme.

Encodes the image device 100 determines as the base parameter QPMBof the quantization parameter QP quantization, which the amount of generated code of the primary coding when coding the input image 91 are predicted to be closest to the target code size based on the size of the generated code low precision, calculated by the encoding and interpolation processing based on the selected parameters QP1 quantization. Encodes the image device 100 encodes (basic coding the input image 91 on the basis of the parameter QPMBquantization by means of block 3 of the main encoding.

Therefore, encodes the image device 100 may determine the baseline parameter QPMBquantization based on the size of the generated code low-precision generated by simple processing, making it possible to simplify the processing of determining basic parameter QPMBthe quantization.

2. The second option exercise

the Second variant of implementation, shown in figure 5-6, is such that the portion corresponding to the first variant implementation, shown in figure 1-4, are denoted by the same symbol, different from the first version of the implementation is that the main encoding unit 3 according codereuse image device 200 performs the feedback control according to the actually generated amount of generated code of the primary encoding.

2-1. The feedback control when the main encoding

Encodes image device 200 in the same manner as in the first embodiment, performs basic coding by feedback control using the basic parameter QPMBquantization, in which the amount of generated code of the primary quantization is predicted as the most approximate to the target code size. Therefore encodes image device 200 can compress the size of generated code of the primary encoding to the target code size. However, occasionally encodes image device 200 is not able to calculate the size of the generated code high precision and consequently chooses the wrong base parameter QPMBthe quantization. Therefore, in order to work and in this case, encodes image device 200 compresses the size of generated by encoding the code to the target code size, the m by means of feedback control of the amount of generated code of the primary encoding. Note that in the second embodiment, the control unit 4 size of the code defines, for example, as the base parameter QPMBof the quantization parameter QP quantization, which the amount of generated code high precision does not exceed the target code size, and is the closest to the target code size. Thus, the control unit 4 size of the code defines the base parameter QPMBquantization so that the size of generated code of the primary coding could be somewhat less than the target code size.

As shown in figure 5, encodes image device 200 performs the following processing for the implementation of high-quality size distribution code in the picture during the execution of the control code in accordance with the compression method of the image represented by the standard H264/AVC (advanced video encoding; improved standard compression / coding of moving images or the like, as well as for the implementation of the encoding of images with a fixed length code.

- When the main encoding three pass encodes image device 200 performs the speed control according vnutrivennoi feedback (Feedback) to compress the code size to a certain value or less.

In order to suppress unwanted fluctuate the parameter QP quantization due to the feedback control, in the case when the size of generated code for each image is predicted as not to exceed the target code size, encodes image device 200 changes the subject to use the parameter QP quantization the quantization parameter) basic parameter QPMBquantization using only visual features (activity). That is, encodes image device 200 takes as its subject the use of the parameter QP for the adaptive quantization parameter QPt quantization with a basic parameter QPMBquantization as the average picture QP.

In the case when the size of the generated code of the primary encoding is predicted to exceed the target code size due to the feedback control, encodes image device 200 modifies the average parameter QP quantization basic parameter QPMBthe quantization. At the same time, to suppress unwanted fluctuations of the average parameter QP quantization encodes image device 200 modifies the average parameter QP quantization only in the direction where the parameter QP quantization becomes large (i.e. only in the direction where the size of the generated code becomes small). The result is that encodes image device 200 takes as the subject ISOE is itaniu parameter QP of the adaptive quantization parameter QPt quantization based on the modified medium setting QP quantization. Note that when the average parameter QP quantization changed more direction, he never returned to the area where he becomes less.

Specifically, the main encoding unit 3 encodes the image device 200 performs encoding using the quantization matrix Q matrix and group activity, found the first predatious unit 1, with the base parameter QPMBquantization, found on the basis of the results of the second predatious block 2, as the average parameter QP quantization. At the same time encodes image device 200 exposes the medium setting QP quantization of the feedback control so that it does not exceed the target code size.

Unit 4 controls the size of the code image coding device 200 performs the feedback control for each increment Edinorosses the feedback control, is composed of a set of macroblocks (MB). The size of generated code output 92 is supplied from block 34 statistical encoding unit 4 controls the size of the code. Unit 4 controls the size of the code uses the size of generated code for each MB (i.e., the output unit 23-1 quantization and block 24-1 length calculations statistical code with the predicted parameter QPd quantization of the second pre-encoding (QP_ is 1) to calculate the target amount of code for each increment Edinorosses the feedback control (hereinafter this will be called "the target code size image) of the target amount of code for each image (hereinafter this will be called the "target size code feedback").

Now, if we say that the size of generated code high-precision increment for the feedback control (Edinorosses) according to the results of the second predatious block 2 equal Predlozheniya.obyazatelno[no], and the size of the generated code high precision for each image according to the results of the second predatious unit 4 equal Predlozheniya, the target code size feedback increment Edinorosses the feedback control (Celebritiescelebrity[no]) is obtained by the following expression. Note that no is the number of Adenocortical (from 0 to Account Edinorosses - 1), and Celebrity is the size of the target code image.

Celebritiescelebrity[no]=

Celebrity×Predlozheniya.obyazatelno[no]/Predlozheniya

That is, the control unit 4 size code multiplies the percentage of the size of generated code high-precision increment for the feedback control (Predlozheniya.obyazatelno[no]) in terms of the size of generated code high accuracy for images on the target code size images Celebrity, making it calculates the size of the generated code feedback (Celebritiescelebrity[no]).

Specifically, the block of the 4 control the size of the code image coding device 200 performs the processing as follows.

(1) Unit 4 controls the size of the code divides the encoding in the first half processing and the second half processing and does not perform the feedback control with the first half of the processing. Unit 4 control code size determines the target amount of code to increment when the feedback control (Adenocortical) according to the size of the code to increment when the feedback control (Adenocortical) pre-encoding. It is the ratio of the size of the code is changed according to the change of the parameter QP quantization. Thus there is a possibility that when performing the feedback control, when the size of the generated code is small in the first half of the coding, encodes image device 200 can change unnecessary medium setting QP quantization. In other words, encodes image device 200 selects the basic parameter QPMBquantization so that the size of generated code of the primary encoding is approaching the target code size of the picture. Accordingly, there is a possibility that when the average parameter QP quantization is rejected at an early stage of the picture, there may be a discrepancy between the size of generated code of the primary encoding and the target code size images.

(2) Block 4 control code size determines the timing of switching between first is albinoi processing and the second half of the processing as from aspects of the stabilization parameter QP quantization, and reduce the size of generated code of the primary encoding to the target code size or less. That is, the control unit 4 size code goes to the second half of the processing when satisfied one of the following conditions in paragraphs (a) and (b).

(a) Unit 4 controls the size of the code compares the available size of the remaining code in the part of the MB, which is not yet encoded (the value obtained by subtracting the size of the generated code of the primary coding up to this point from the target code size images), and the target code size in terms of MB, which is not yet encoded (the value obtained by subtracting the target code size feedback up to this point from the target code size images). When the size of the remaining code a certain percentage or less unit 4 controls the size of the code goes to the second half of the processing. This is to reduce the size of generated code of the primary encoding to the target code size or less.

Condition (2)(b) represented by the following expression.

(Telebit)-Genericlisinopr<

(Telebit-Celebritycouple)X Dorogovato

Here Celebritycouple and Genericlisinopr represent respectively the integer value of the target code size feedback up to this point (Celebritiesprivate[no]) and a is isleno the value of the size of generated code of the primary encoding increment when the feedback control up to this point. In addition, Dorogovato is a tolerance percentage. In the case when the size of the generated code of the primary encoding is equal to the target code size, the size of the remaining code is equal to the target amount of code in the part which is not yet encoded. On the other hand, in the case when the size of the generated code of the primary encoding does not exceed the target code size, the size of the remaining code is less than the target amount of code in the part which is not yet encoded. For example, when the size of the remaining code reaches less than Dorogovato time from the target code size in the part which is not yet encoded, block 4 control code size is transferred to the second half of treatment. In the case of performing control so that the size of the generated code must be less than the target code size, unit 4 controls the size of the code specifies Dorogovato to a value less than 1. Thus, when the size of the generated code in the part that is encoded is less than the target code size, and reaches the target code size, unit 4 controls the size of the code goes to the second half of the processing. In addition, in the case of performing control so that the size of the generated code is approaching the target code size, unit 4 controls the size of the code specifies Dorogovato the value is 1 more. Thus, when the size of the generated code in the part that is encoded exceeds the target code size with a certain ratio or more, the control unit 4 size of the code can move to the second half of the processing. This happens without that Dorogovato can be set to "1".

Thus, the unit 4 control code size can enter the feedback control before you receive the difference between the target code size and the real size of the generated code of the primary encoding, and therefore, can prevent a situation in which the size of the remaining code is so small that the unit 4 control code size cannot control the size of the generated code of the primary encoding properly.

In other words, although the size of generated code of the primary encoding is small, the absolute value of the size of the rest of the code is large, and accordingly, the condition is not satisfied as long as the difference between the target code size and the size of generated code of the primary encoding does not become too large, the control unit 4 size of the code does not enter the second half of the processing for entry into the feedback control. When the size of the generated code of the primary encoding increases, the absolute value of the size of the remaining code a little, and, accordingly, slowie satisfied even if there is any discrepancy between the target code size and the size of generated code of the primary encoding, unit 4 controls the size of the code goes to the second half of the processing for entry into the feedback control.

(3) Block 4 control code size changes the medium setting QP quantization in the direction to increase to become greater than the baseline parameter QPMBquantization, only in the case of predicting that the size of generated code of the primary encoding for each image will exceed the target code size of the picture in the second half of treatment. In the case of an increase in the average parameter QP quantization unit 4 controls the amount of code changes the medium setting QP quantization on 1 pay processing with feedback (i.e. for each increment when the feedback control). Thus, the control unit 4 size code suppresses the excessive change in the average parameter QP quantization.

In the case of fitting to one condition of the following (a) and (b) unit 4 controls the size of the code predicts that the size of generated code of the primary encoding for each image will exceed the target code size images, and increases the average parameter Base QP quantization.

(a) Unit 4 controls the size of the code confirms the size of generated code feedback before (immediately after the end of treatment), as of the end of the encoding prires the Institute during the feedback control, and compares the size of the generated code with feedback (Generalitiespanel[cur]) and the size of the code with feedback (Telefonicadevelopedby.html[cur]) in accordance with the following expression.

Generalitiespanel[cur]>Celebritiesprivate[current]

When the size of the generated code feedback (Generalitiespanel[cur]) is larger than the target code size feedback (Celebritiesprivate[cur]), there is a tendency in which the size of generated code increases, and, accordingly, this gives an indication that there is a possibility the size of generated code of the main encoding generated in the subsequent encoding to exceed the target code size. At the same time, the control unit 4 size code determines whether to increase the average parameter QP quantization. Condition at the same time will be shown later.

Unit 4 controls the size of the code subtracts the size of the generated code from the target amount of code in the part that is encoded at the same time, making it calculates the remaining amount of generated code Spitbite as follows.

Isbitset=Celebritycouple-Genericlisinopr

Unit 4 control code size increases the average parameter QP quantization only in the case of predicting that the size of generated what about the code of the primary coding in the next increment when the feedback control will exceed the target code size. Unit 4 controls the size of the code retrieves the maximum predicted code size to exceed the target code size in the next increment for the feedback control (hereinafter this will be called "excess the maximum size of the code") of the target code size in the next increment for the feedback control in the following way. Here Makatsoris represents the maximum ratio that can take the size of generated code feedback as errors in relation to the target code size feedback.

Celebritiesprivate[next]×Makatsoris

Unit 4 controls the size of the code executes based on the following definition.

Spitbite<Celebritiesprivate[next]×Makatsoris

Unit 4 controls the size of the code compares the balance of the size of the generated code Spitbite and excessive the maximum size of the code in the next increment for the feedback control. In the case where excessive the maximum code size is larger than the remaining size of generated code Spitbite, block 4 control code determines that there is the possibility that the size of generated code of the primary encoding part which is encoded may exceed the target code size in the next increment control when the feedback, and gives the increment of the average parameter QP quantization 1.

That is, when the size of the generated code in the last increment when the feedback control is larger than the target code size of the last increment when the feedback control", and "the rest of the code up to this point Spitbite less than excessive the maximum size of the code in respect of which there is a possibility to exceed the following increment for the feedback control, the control unit 4 size code gives the increment of the average parameter QP quantization 1. This is because there is a tendency for the size of the generated code to increase, and insufficient allowance for the remainder of the code size, and, accordingly, is the prediction that the final size of the generated code of the picture exceeds the target code size images.

In other words, when determining that the size of generated code of the primary encoding part that is encoded, will exceed the target code size in the next increment when the feedback control even when supported by the current state, the control unit 4 size code does not increment the average parameter QP quantization 1. Thus, even when there are indications that the size of generated code of the primary encoding part which Secadero the Ana, may exceed the target code size, block 4 control code size can avoid over-reaction, to prevent excessive unnecessary increase in the average parameter QP quantization and stabilize the average parameter QP quantization.

(b) In the case of non-compliance (a), i.e. when the size of the generated code in the last increment when the feedback control is smaller than the target code size of the last increment when the feedback control or the rest of the code up to this point Spitbite more than excessive the maximum size of the code that has the ability to exceed in the next increment for the feedback control, the control unit 4 size code confirms if a negative balance of the size of generated code Spitbite. When Spitbite negative, this means that the size of generated code of the primary encoding when the encoding ends, exceeds the target code size. At this time, the control unit 4 size code predicts the deviation of the size of generated code in the remaining encoded portion from the target on the basis of the ratio between the target code size and the size of generated code in the last increment when the feedback control as follows.

Unit 4 controls the size of the code calculates the rate between the target code size and the size of generated code in the last increment when the feedback control Rasnita as follows.

Rasnita=(Generalitiespanel[current]-Celebritiesprivate[current])/Celebritiesprivate[current]

Unit 4 control calculates a target code size the code size of the remaining coded part Televista as follows.

Televista=Celebrit-Celebritycouple

Unit 4 controls the size of the code assumes that the target code size of the remaining coded part is rejected with the same ratio as the ratio between the target code size and the size of generated code in the last increment when the feedback control Rasnita, and calculates the deviation from the target size, as follows.

Televista*Rasnita

Unit 4 controls the size of the code gives the increment of the average parameter QP quantization to 1 when the result obtained by adding the coded part and a part that is not yet encoded, is positive and satisfies the following condition.

(Televista*Rasnita)-Spitbite>0

When the size of generated code encoded portion exceeds the target code size", and "predicted size of the generated code image obtained from the ratio between the target code size and the size of generated code in the last increment when the feedback control exceeds the target amount of code of the picture unit 4 controls the size of the code gives the increment of the average parameter QP quantization 1.

That is, even when the size of the generated code at some point exceeds the target code size, if it is determined that the size of generated code is declining in relation to the target code size, and thus, the size of the generated code image does not exceed the target code size images, block 4 control code size will not exceed the average parameter QP quantization 1.

Thus, when the input image 91 is exposed to the basic encoding, encodes image device 200 does not perform the feedback control during the beginning of the main coding for images, and performs the feedback control for the second half of treatment, only when a certain condition is satisfied.

That is, encodes image device 200 performs almost the feedback control only for the second half of the processing. Even in the first half of the processing in the case where the ratio between the size of the generated code encryption part and the target code size exceeds a certain value or more, there is the possibility of divergence between the size of the subsequently generated code and the target code size, and, therefore, encodes the image device 200 performs the feedback control.

Encodes the image in trojstvo 200 suppresses fluctuations of the average parameter QP quantization to a minimum when the feedback control, and, accordingly, only when it is predicted that the size of generated code of the primary encoding for each image is not suppressed immediately to the target code size or less, gives the increment of the average parameter QP quantization 1.

That is, in the case when the size of the generated code feedback is greater than the target code size of the feedback, there is a tendency for the size of the generated code is increased compared with the target code size, and if the size of the generated code encryption unit may exceed the target code size in the next increment for the feedback control, is the prediction that the final size of the generated code of the picture exceeds the size of generated code of the picture and, therefore, encodes the image device 200 provides an increment to the average parameter QP quantization 1.

Thus, even in the case when there is a tendency for the size of the generated code is increased compared with the target code size, if there is allowance for the size of the remaining code, encodes the image device 200 can prevent the build-up 1 medium setting QP quantization and, accordingly, prevents an unnecessary increase in the average parameter QP quantization.

Further, in the case when the medium is a QP quantization has not received increment by 1 in the case when the size of the generated code encoded portion exceeds the target code size and makes the prediction that the size of generated code end images will exceed the target code size images, encodes image device 200 provides an increment to the average parameter QP quantization 1.

Thus, even in the case when the size of the generated code encoded portion exceeds the target code size, and in the case when the size of the generated code tends to decrease and makes the prediction that the size of generated code of the picture exceeds the target code size images, encodes image device 200 can prevent the increment of the average parameter QP quantization 1 and, accordingly, prevents an unnecessary increase in the average parameter QP quantization.

2-2. Procedure

Procedure RT4 processing of the feedback control will be described with reference to the block diagram of algorithm 6. At the beginning of this processing unit 4 controls the size of the code specifies the base parameter QPMBquantization defined by predatirovaniya, on medium setting QP quantization (step S31). After that, the control unit 4 size code encodes the input image 91 for each increment when the feedback control of Adenocortical (step S32).

Then b is OK 4 control code size determines encoded all MB image (step S33). Here, in the case when all of the MB encoded, block 4 control code moves to the end step at the end of processing. On the other hand, in the case when all the MB is not coded, unit 4 controls the size of the code proceeds to the next step S34. Unit 4 control code size determines coded whether a certain number of MB or more (step S34). That is, as described above, there is a possibility that when the feedback control performed when the code size is small in the first half of treatment, the medium setting QP quantization can be changed unnecessarily.

Here, when it is determined that encoded a certain number of MB or more, the unit 4 control code size goes to step S36. On the other hand, when it is determined that a certain number of MB or more not coded, unit 4 control code size goes to step S35, compares the available size of the remaining code in the part of the MB, which is not yet coded (code size obtained by subtracting the size of the generated code up to this point from the target code size), and the target code size in terms of MB, which is not yet encoded, and determines, is whether a certain percentage or less, the ratio between the size of the remaining code and the target code size in the part which has not yet Zack is derovan (step S35). Here, when it is determined that the ratio between the size of the rest of the code and the target code size in the part which is not yet encoded, it is not certain percentage or less, the macroblock, which is now in process of coding is not the second part of the processing, and there is a discrepancy between the size of the generated code encryption part and the target code size, and, accordingly, the control unit 4 size code is returned to step S32 to repeat the above processing. On the other hand, when it is determined that the available code size is a certain percentage or less, the unit 4 control code size goes to step S36 to perform the feedback control.

Following this, the control unit 4 size code determines whether the size of the generated code in the last increment for the feedback control of the target code size in the last increment when the feedback control (step S36). Here, when it is determined that the size of generated code in the last increment when the feedback control is larger than the target code size in the last increment for the feedback control, the control unit 4 size code determines whether there is a possibility that the rest of the code up to this point can be higher than the last increment when the feedback control (step S37). In the case when the size of the generated code in the last increment when the feedback control is smaller than the target code size in the last increment for the feedback control, the control unit 4 size code goes to step S39.

When it is determined that the rest of the code up to this point may be exceeded in the next increment for the feedback control, the control unit 4 size code gives the increment of the average parameter QP quantization to 1 (step S38) and returns to step S32. On the other hand, when it is determined that the rest of the code up to this point may not be exceeded in the next increment for the feedback control, the control unit 4 size code goes to step S39.

After that, at step S39, the control unit 4 size code determines whether the size of generated code encoded in the target code size (step S39). Here, when it is determined that the size of generated code coded part does not exceed the target code size, block 4 control code size is returned to step S32 to repeat the above processing.

On the other hand, when it is determined that the size of generated code encoded portion exceeds the target code size, block 4 control code size determines whether predskazan is th size of the generated code image, obtained from the ratio between the target amount of code and generated code size in the last increment for the feedback control, the target size of the code image (step S40). This was followed in the case when it is determined that the predicted size of the generated code image does not exceed the target code size images, block 4 control code size is returned to step S32 to repeat the above processing.

When it is determined that the predicted amount of generated code of the picture exceeds the target code size images, the control unit 4 size code gives the increment of the average parameter QP quantization to 1 (step S38) and returns to step S32 to repeat the above processing. Thus, the control unit 4 size code continues this process up until the step S33 is not determined that all of the MB encoded.

As described above, according to the present invention encodes image device 200 may encode the image with a fixed length code without using binary search, and, consequently, the size of circuits and the power consumption can be reduced, and the latency of the encoding can be reduced. At the same time encodes image device 200 suppresses unwanted fluctuations of the average parameter QP quantization due to control reverse is th link, and, accordingly, even when embodied the feedback control can be optimal distribution of code size, i.e. the size distribution of the code to ensure that the size of generated code of the picture coincides with the target code size pictures, and taking into account the visual characteristics.

Note that the present invention is not limited to the above embodiment, and various modifications or changes without departing from the essence of the present invention.

For example, the above encodes the image device 100 and a method of coding images can also be implemented as a computer program installed in the device, the storage medium on which is recorded the program, like a computer program that implements the method, or a storage medium on which is recorded the program.

The above processing sequence encoding may be performed by hardware and can also be performed programmatically. In the case of execution of the processing sequence encoding software encodes image device 200 is formed in virtual CPU and RAM. After this encoding program stored in ROM, is overwritten in the RAM, and executes the processing of the encoding.

2-3. Work and benefits

In the above embodiment, the OS is enforced encodes image device 200 determines the baseline parameter QP MBquantization serving as the quantization parameter in respect of which predicts that the size of generated code of the primary encoding for each picture when coding the input image 91 is closer to the target code size images.

Encodes image device 200 encodes the input image 91 for each increment when the feedback control by performing quantization with adaptive parameter QPt quantization based on the average parameter QP quantization, which serves as the used quantization parameter found on the basis of at least the basic parameter QPMBthe quantization.

Encodes image device 200 confirms the size of generated code of the encoded input image 91 for each increment in the feedback control in the case of predicting that the size of generated code for each increment when the feedback control will exceed the target code size for each increment of the image, increases the average parameter QP quantization, thereby increasing the value of the adaptive parameter QPt quantization.

Thus, encodes image device 200 limited increases the average parameter QP quantization in a state in which the size of GE is erasemode code of the primary encoding is regulated in advance to get closer to the target code size of the picture, thus you can protect the value of the subject using the adaptive parameter QPt quantization of unnecessary deviations.

In the encoding method, followed by quantization, such as AVC or the like, confirmed the deterioration of image quality due to an executable lot of time coding. To prevent this deterioration in image quality is proposed a method called reverse tracing, which is determined by the parameter QP quantization used during the last encoding, and is used again (for example, see Patent document 2).

Patent document 2: international application PCT/JP 2008/066917

When the reverse tracking detection range tracing is set based on the predicted parameter QPt quantization, which can simplify the processing limit of the range tracking in the backward tracking.

Encodes image device 200 is able to suppress the fluctuations to be used parameter QP quantization, and, accordingly, subject to the use of the parameter QP quantization can be protected from deviation from the range tracking in the backward tracking, and detection speed in the reverse tracking can predochranitel deterioration.

If there is a tendency for the size of the generated code feedback for each increment when the feedback control increases and there is an insufficient allowance for the remainder of the code size obtained by subtracting the size of the generated code encoded part of the target code size coded parts in the picture, which is the increment image, encodes the image device 200 predicts that the size of generated code of the picture exceeds the target code size images.

When the size of the generated code in the last increment when the feedback control is larger than the target code size in the last increment for feedback control, encodes image device 200 determines that there is a tendency, which increases the size of generated code for each increment in the feedback control.

When the size of the generated code in the next increment when the feedback control is less than the excessive the maximum size of the code in respect of which there is a possibility that it will exceed the target code size, encodes image device 200 determines that there is insufficient allowance for the remainder of the code size.

Thus, encodes image device 200 increases the average parameter is QP quantization only in a limited case, when there is a tendency, which increases the size of generated code feedback and there is an insufficient allowance for the remainder of the code size, whereby it is possible to prevent unnecessary fluctuations of the average parameter QP quantization.

Assuming that the size of generated code encoded portion of the picture exceeds the target code size of this encrypted part, and that the ratio between the target code size and the size of generated code does not change when the size of generated code of the picture exceeds the target code size images, encodes image device 200 predicts that the size of generated code of the picture exceeds the target code size images.

In the case when the predicted value of the amount of generated code of the picture obtained from the ratio between the target code size of the feedback and the size of generated code feedback in the last increment when the feedback control (the predicted size of generated code of the picture)exceeds the target code size under the assumption that the ratio of the target code size and the size of the generated code has not changed, encodes image device 200 determines that the size of generated code of the picture exceeds the target code size images.

Thus, encodes image is agenia device 200 increases the average parameter QP quantization only in the limited case, when the size of the generated code of the picture exceeds the target code size of the picture when supported by the ratio between the size of the generated code feedback and the current state of the target code size feedback, whereby it is possible to prevent unnecessary fluctuations of the average parameter QP quantization.

Encodes image device 200 predicts will exceed it in the second half of the picture size of the generated code of the picture target code size of the picture. Thus, encodes image device 200 is not a performs feedback control in the first half of the processing, where there are many remaining parts for further coding and fluctuations of the average parameter QP quantization have a large impact on the size of the generated code image. The result is that encodes image device 200 can prevent the imbalance of the amount of generated code of the picture due to unnecessary deviations of the average parameter QP quantization in the first half of the processing.

In the case when there is a discrepancy between the size of the generated code encoded part image and the target code size of this encoding part encodes the image device 200 predicts will exceed if the size of generated code kartenkatalog code size pictures, regardless of the position of the image.

In the case when the size of the generated code encoded portion of the picture is smaller than the value obtained by multiplying the target amount of code for this encrypted part on the amount of allowable deviation Dorogovato, encodes image device 200 determines that there is a discrepancy between the size of the generated code image and the target code size of this encryption.

Thus, for example, in the case of a failure in the evaluation of the underlying parameter QPMBquantization encodes image device 200 may perform the feedback control even at an early stage pictures, through which you can handle unexpected situations.

In the case of predicting that the size of generated code of the picture exceeds the target code size images, encodes image device 200 provides an increment to the average parameter QP quantization 1.

Thus, encodes image device 200 can suppress the fluctuations of the average parameter QP quantization to a minimum and to stabilize the image quality in the image.

Encodes image device 200 determines the baseline parameter QPMBquantization so that the size of the generated code when coding the input image 91 is smaller than the target code size is.

Thus, encodes image device 200 can compress the size of generated code images, allowing significantly easier to manage, so the size of the generated code image does not exceed the target code size images.

Encodes image device 200 sets a value less than 1 as the size of the permissible deviation of Dorogovato. Thus, in the case when the size of the generated code encoded part is smaller than the target code size of this encryption part, but close to it, encodes image device 200 may perform the feedback control, whereby control can be performed properly, so that the size of generated code image does not exceed the target code size images.

Encodes image device 200 determines a predicted parameter QPd quantization predicted for the approximation to the underlying parameter QPMBquantization based on the size of the generated code calculated by the encoding mentioned input image 91 by using the parameter QP quantization of the selected parameter QP1 quantization), covering a wide range, and defines the base parameter QPMBquantization based on the size of the generated code high precision, calculated by coding the input image 91 by using this pre is said parameter QP1 quantization parameter QP quantization near the predicted parameter QPd quantization.

Thus, encodes image device 200 determines the baseline parameter QPMBquantization with high accuracy, whereby the amount of generated code of the primary quantization can be approximated to the target code size images in this basic parameter QPMBthe quantization. Accordingly, encodes image device 200 only occasionally will have a medium setting QP quantization flucturion due to the feedback control, whereby the fluctuations of the average parameter QP quantization can be suppressed to the maximum extent.

According to the above variant implementation encodes image device 200 performs the feedback control so as to increase the average parameter QP quantization only in the case of predicting that the size of generated code of the picture exceeds the target code size of the picture, the size of the generated code encryption part and the target code size of this encryption.

Thus, encodes image device 200 may reduce the size of generated code image to the target code size images or less in the right way during the suppression of the fluctuations of the average parameter QPMBquantization to a minimum. Thus, the present invention can realize the us is a device and image processing method, image processing in which the amount of generated code of the picture can be reduced to the target code size images or less correctly while maintaining uniform quality of the image.

3. Other embodiments of the

Note that in the above second embodiment described the case of three-pass configuration, which encodes image device 200 determines a predicted parameter QPd quantization on the basis of the results of the first predatious block 1 and defines the base parameter QPMBquantization based on the results of the second predatious unit 2. The present invention is not limited to this and, for example, the present invention can be applied to codereuse image device, implemented in a two-pass configuration in which the first precederei block 1 defines the base parameter QPMBthe pre-encoding, or four or more loop-through configuration.

In addition, in the first and second of the above embodiments perform the described case in which the input image 91 is encoded vnutrikarernym prediction, orthogonal transformation according to DCT, quantization and coding according to CAVLC or SAVAS. The present invention is not limited to this, and any of them can be omitted if the input image 91 is exposed on men is her least quantization. In addition, the input image 91 can be encoded using an encoding method other than these.

Further, in the above-described second embodiment described a case in which the size of the generated code feedback confirmed for each increment in the feedback control, composed of a set of MB. The present invention is not limited to this, and the size of the increments used when the feedback control is not limited, and, for example, the increment for the feedback control can be set for MB or part.

Further, in the above-described second embodiment describes the case in which the size of generated code for each image, serving as increments of the image is reduced to the target code size or less. The present invention is not limited to this and, for example, the size of generated code for each picture or for each set of images that serve as the increment of the image can be reduced to the target code size or less.

Further, in the above-described second embodiment describes the case in which the medium setting QP quantization incremented by 1 with respect to the feedback control as a single increment for feedback control. The present invention is not limited to this, and the average parameter Base is QP quantization can get an increment is two or more, depending on the conditions.

Further, in the above-described first and second embodiments perform the described case in which, if the latter is the size of generated code feedback exceeds the target code size of the feedback, it is determining that there is a tendency for the size of generated code for each increment when the feedback control is increased. The present invention is not limited to this and, for example, it may be determined that there is a tendency for the size of generated code for each increment when the feedback control is increased depending on whether the size of the generated code to the target code size for each of the last set of increments when the feedback control.

Further, in the above-described second embodiment described a case in which when getting under one of the conditions (3)(a) and (3)(b) predicts that the size of generated code for each increment of the image exceeds the target amount of code for the above each increment of the image. The present invention is not limited to this, and you can make the prediction that the size of generated code for each increment of the image exceeds the target code size for each of the above increment image in the case falling under any of them, or the other method.

Further, in the above-described second embodiment described a case in which prediction is about to exceed whether the size of the generated code of the picture target code size of the picture, the ratio between the final size of generated code feedback and target size code feedback. The present invention is not limited to this and, for example, you can make a prediction about what will exceed if the size of generated code of the picture target code size, the ratio between the target code size and the size of generated code for each set of increments when the feedback control or for each MB.

Further, in the above-described second embodiment described a case in which when entering the second half of processing images, the feedback control is performed automatically. The present invention is not limited to this and, for example, the feedback control can be performed only in the case when the size of the generated code encoded portion exceeds the target code size.

Further, in the above-described second embodiment describes the case in which the feedback control is performed if the size of generated code encoded portion exceeds the target code size. The present invention is not limited to this, and the management of reverse vazuojam be performed according to other various types of conditions.

Further, in the above-described second embodiment describes the case in which the base parameter QPMBquantization is determined so that the size of the generated Corda main coding is slightly less than the target code size of the picture. The present invention is not limited to this basic parameter QPMBquantization can be determined so that the size of generated code of the primary encoding is closest to the size of the generated code image.

Further, in the above-described second embodiment describes the case in which the present invention is applied to AVC. The present invention is not limited to this, the present invention can be applied to various types of encoding for adaptive VLC table selection. For example, in the case where the present invention is applied to MPEG-2, as the quantization factor is used to scale the quantization.

Further, in the above-described second embodiment described a case in which encodes image device 200 serving as the imaging device is composed of first predatious block 1, the second predatious block 2 and block 4 control code size, serving as blocks determine the underlying factors of quantization, the main encoding unit 3 serving as co is youseo block, and unit 4 controls the size of the code, which serves as a control unit with feedback. The present invention is not limited to this, and the imaging device according to the present invention may be composed of blocks determine the underlying factors of quantization according to other types of configurations, the coding unit and control unit with feedback.

1. The imaging device containing:
the definition block of the base quantization factor, made with the ability to determine the underlying factor of quantization, which is predicted so that the size of generated code for each increment of the image when encoding the input image will be approximated to the target code size for each increment of the image;
the coding block, configured to encode the input image for each increment when the feedback control by performing quantization using quantization factor, defined on the basis of at least the basic quantization factor; and
the power feedback control is performed with the opportunity to confirm the size of generated code for the input image coded by the coding block for each increment in the feedback control, and in the case of predicting what time is ur code to be generated for each increment of the image will be larger than the target code for each increment of the image, to increase the used quantization factor, which is
the above-mentioned control unit with feedback predicts that the size of generated code for the above-mentioned each increment of the image exceeds the target code size for these each increment of the image when the size of generated code for each referred to increment when the feedback control tends to increase, and there is an insufficient allowance for the size of the rest of the code obtained by subtracting the size of the generated code part, coded from the target code size part, encrypted with said each increment of the image.

2. The imaging device according to claim 1 in which the said control unit with feedback determines that the size of generated code for the above-mentioned each increment when the feedback control tends to increase when the size of the generated code of the last increment when the feedback control is larger than the target code size of the last increment in the feedback control.

3. The imaging device according to claim 2, in which the said control unit with feedback determines that there is an insufficient allowance for the above-mentioned size of the remaining code, when referred to the size of the remaining code is less than the superfluous is s the maximum size of the code, in respect of which there is a possibility that the size of generated code with the following increment when the feedback control may exceed the target code size.

4. The imaging device according to claim 1 in which the said control unit with feedback predicts that the size of generated code for each increment of the image exceeds the target code size for these each increment of the image when the size of generated code part, encrypted with the aforementioned increment image exceeds the target code size of this encrypted part, and the size of generated code for each increment is larger than the target code size for these each increment of the image under the assumption that the ratio between the target code size and the size of the generated code is not changed.

5. The imaging device according to claim 4 in which the said control unit with feedback determines that the size of generated code for each increment is larger than the target code size for these each increment of the image under the assumption that the ratio between the said target amount of code and the generated code size is not changed, when the predicted value of the size of generated code for each UPR is analogo increment image, obtained from the ratio between the target code size and the size of generated code of the last increment when the feedback control exceeds the target code size.

6. The imaging device according to claim 5 in which the said control unit with feedback predicts will exceed if the size of generated code for the above-mentioned each increment of images of the target code size for these each increment of the image in the last half of the mentioned units of the image.

7. The imaging device according to claim 6, in which the said control unit with feedback predicts will exceed if the size of generated code for the above-mentioned each increment of images of the target code size for these each increment of the image regardless of the position referred to increment the image in the case where there is a discrepancy between the size of the generated code part, encrypted with the said unit image, and the target code size of this encryption.

8. The imaging device according to claim 1 in which the said control unit with feedback determines that there is a discrepancy between the size of the generated code coded part with the said increment image and the target code size coded in this part in the case, to the Yes, the size of generated code part, encoded with the above-mentioned increment image is smaller than the value obtained by multiplying the amount of deviation of the target code size of this encryption.

9. The imaging device according to claim 1 in which the said control unit with feedback increases mentioned used the quantization factor of one in the case of predicting that the size of generated code for the above-mentioned each increment of the image exceeds the target code size for these each increment of the image.

10. The imaging device of claim 8 in which the said control unit with feedback determines the mentioned base quantization factor so as to reduce the size of generated code for encoding the mentioned input image is less than said target code size.

11. The imaging device of claim 10 in which the said value of the permissible deviation is obtained from the value that is lesser than 1.

12. The imaging device according to claim 1 in which the said block encoding accept-mentioned basic quantization factor defined for the above-mentioned each increment of the image, the average quantization factor and uses the value to which is added the quality factor of the quantization shift? what about the activity that the average quantization factor; while the above-mentioned control unit with feedback increases mentioned used the quantization factor by increasing the mentioned average quantization factor.

13. The imaging device according to claim 6, in which the said block defining a base quantization factor determines the predicted quantization factor, which predicts that it is close to the base quantization factor based on the size of the generated code calculated by the encoding mentioned input image using the quantization factor, covering a wide range, and determines the mentioned base quantization factor based on the size of the generated code calculated by the encoding mentioned input image with that of the predicted quantization factor and the quantization factor near the predicted quantization factor.

14. The method of image processing, comprising stages, which are:
define the basic quantization factor to determine the basic quantization factor that is associated with the prediction that the size of generated code for each increment of the image when encoding the input image will be approximated to the target size of generated code for each increment of the image;
encode the input image for each ol the treatment when the feedback control, to generate the coded stream by performing quantization using the used quantization factor, defined on the basis of at least the basic quantization factor; and
control with feedback to confirm the size of generated code for the input image coded by the coding block for each increment in the feedback control, and in the case of predicting that the size of generated code for each increment of the image will be larger than the target code for each increment of the image, used to increase the factor of quantization, in which
during feedback control predict that the size of generated code for the above-mentioned each increment of the image exceeds the target code size for these each increment of the image when the size of generated code for each referred to increment when the feedback control tends to increase, and there is an insufficient allowance for the size of the rest of the code obtained by subtracting the size of the generated code part, coded from the target code size part, encrypted with said each increment of the image.



 

Same patents:

FIELD: physics, communication.

SUBSTANCE: invention relates to picture digital signal encoder/decoder in appropriate chroma signal format. Encoder comprises module to divide input bit flow into appropriate colour components, module to divide colour component input signal into blocks so that coding unit area signal can be generated, module to generate P-frame for said signal, module to determine prediction mode used for encoding by efficiency P-frame prediction, module to encode prediction error for encoding difference between predicted frame corresponding to prediction mode as defined by appropriate module and colour component input signal, and module for encoding with variable length of prediction mode code, output signal of prediction error encoding module and colour component identification flag indicating colour component wherein belongs said input bit flow resulted from division of colour components.

EFFECT: higher efficiency.

2 cl, 98 dwg

FIELD: information technology.

SUBSTANCE: disclosed is a liquid crystal display device comprising: means (14) of converting frame frequency via interpolation of an image which is corrected based on an interframe motion vector between frames of input video signals. During interpolation of signals which are represented by two or three successive identical images generated in television projector equipment for use in the next interpolated image, motion vectors are recalculated based on motion vectors (Va) used in generating the first interpolated image, thereby identifying motion vectors having high accuracy and fewer errors; a unit (10) for determining a repeated image, and uses an interval during which identical images successively appear without the need for identifying vectors for further increase in accuracy of motion vectors identified between the image and the other immediately preceding image.

EFFECT: high accuracy of identified vectors and generating interpolated images of higher quality, even if the device processes such data as three to two pull-down or two to two pull-down, having high probability of deterioration of accuracy of identified vectors due to a wide interval of movement.

3 cl, 6 dwg

FIELD: information technology.

SUBSTANCE: disclosed is a method for scalable encoding of video information, which calculates a weighting coefficient and indicates change in brightness between the encoded target region of the image and the region of a reference image in an overlying layer, calculates a motion vector, and generates a prediction signal by applying the weighting coefficient to the decoded signal of the region of the reference image indicated by the motion vector, and compensating for movement. If the immediate underlying region of the image has performed interframe prediction in the immediate underlying layer, the method identifies the region of the reference image of the immediate underlying layer which was used by the immediate underlying region of the image as a prediction reference for predicting movement, and calculates a weighting coefficient by applying the weighting coefficient which was used by the immediate underlying region of the image in weighted prediction of movement, to the DC component of the region of the image in the overlying layer, and accepting the result of application as the DC component of the immediate underlying region of the image.

EFFECT: high efficiency of scalable encoding.

26 cl, 24 dwg, 2 tbl

FIELD: information technology.

SUBSTANCE: target frame of a reference vector and a reference vector from already encoded frames are selected; information for labelling each frame is encoded; the reference vector is set to indicate a region in the target frame of the reference vector relative the target encoding region; the reference vector is encoded; the corresponding regions are searched using image information of the target region of the reference vector belonging to the target frame of the reference vector and is indicated through the reference vector and the reference frame; the reference region in the reference frame is determined based on the search result; a predicted image is formed using reference frame image information corresponding to the reference region; and the difference information between the target encoding region information and the predicted image is encoded.

EFFECT: efficient encoding of vector information used for inter-frame predictive coding, even when the reference frame used in the inter-frame predictive coding differs between the target encoding region and its neighbouring region.

24 cl, 17 dwg

FIELD: information technologies.

SUBSTANCE: method is proposed to code images, in which a coding object pixel value is forecasted, and the forecasted value is produced; data of probability distribution is calculated, which indicates the value that the initial pixel value has for the produced predicted value, by means of shifting, according to the predicted value, data of differential distribution of difference between the initial value of the pixel and the forecasted value when coding with forecasting. Data of differential distribution are stored in advance; the produced data of probability distribution are cut-off to hold data in the range from the lower limit to the upper limit for possible values of the initial pixel value; and the coding object pixel value is coded using cut-off data of probability distribution of the initial pixel value from the lower limit to the upper limit.

EFFECT: higher efficiency of coding with forecasting by rejecting calculation of difference between an initial pixel value and its forecasted value when doing time and space forecasting.

10 cl, 19 dwg

FIELD: information technology.

SUBSTANCE: videocoding method is suggested to be formed on the basis of information about mismatch between already coded reference image of camera and being coded target image of camera. This method includes step in which for each predetermined unit sections in differential image of one of following groups are selected: group of decoded differential image obtained using decoded differential image between already coded image of camera and image with compensated discrepancy, and group of decoded camera image obtained using decoding already coded camera image by means of determination if there is image with compensated discrepancy in corresponding position or not, i.e. if corresponding pixel in the image with compensated discrepancy has effective value or not.

EFFECT: higher efficiency of coding video with multiple viewpoints applying movement compensation to differential image, lower prediction difference in the part with both time redundancy and redundancy between cameras.

14 cl, 9 dwg

FIELD: information technology.

SUBSTANCE: image processing device is suggested including acquisition module intended to obtain data of moving image, containing multiple sequential frames and one or more image data corresponding to the frames and having higher spatial resolution then these frames; movement prediction module intended to detect movement vector between frames using moving image data; difference value calculation module intended to calculate difference value between given frame and frame corresponding to image data; and image generation module which allows to generate image data with compensated movement which data corresponds to given frame, on the basis of frame corresponding to image data and movement vector.

EFFECT: prevention of noise interference in high-frequency image data component during data generation for image with high spatial resolution by means of movement prediction on the basis of data sequence of image with low spatial resolution and movement compensation using data of image with high spatial resolution.

10 cl, 7 dwg

FIELD: information technology.

SUBSTANCE: image processing device is suggested including acquisition module intended to obtain data of moving image, containing multiple sequential frames and one or more image data corresponding to the frames and having higher spatial resolution then the frames; movement prediction module intended to detect movement vector between frames using moving image data; and image generation module intended to generate image data with compensated movement which data corresponds to given frame, on the basis of image data and movement vector. Image generation module generates data of image with compensated movement which data is located between certain frame and frame corresponding to image data and corresponding to frame and generates data of image with compensated movement which data corresponds to given frame, on the basis of data of image with compensated movement and movement vector.

EFFECT: movement compensation with high accuracy.

10 cl, 7 dwg

FIELD: information technology.

SUBSTANCE: apparatus for encoding video signal generates a synthetic image for a camera used to obtain a target encoding image using an already encoded reference image of the camera, having a viewpoint different from the viewpoint of the camera used to obtain the target encoding image, and information on disparity between the reference image of the camera and the target encoding image, thereby encoding the target encoding image. A predicted image is formed for a difference image between the input image of the target encoding region which needs to be encoded and the formed synthetic image, and a predicted image is formed for the target encoding region, which is represented by the sum of the predicted difference image and the synthetic image for the target encoding region. The prediction residue is encoded, represented by the difference between the predicted image for the target encoding region and the target encoding image in the target encoding region.

EFFECT: providing a vide encoder and decoder in which there is no need to process different bit depths by forming a predicted image for the input image.

28 cl, 14 dwg

FIELD: information technology.

SUBSTANCE: invention proposes to process video images by applying a time or spatial interframe prediction encoding to each area of the divided image and generating a predicted image of a processing target area based on a reference frame of the processing target area and reference information which indicates a predicted target position of the processing target area in the reference frame, and the predicted reference information is generated as predicted information of the reference information. Reference information used when processing the target area neighbouring the processing target area is defined as prediction data of the predicted reference information used when predicting reference information of the processing target area. Reference information of the reference area is generated using one or more parts of the reference information used when processing the reference area indicated by prediction data. Prediction data of the predicted reference information are updated using reference information of the reference area. Predicted reference information is generated using or more parts of prediction data updated by predicted reference information.

EFFECT: high efficiency of encoding information for interframe prediction.

25 cl, 29 dwg

FIELD: video decoders; measurement engineering; TV communication.

SUBSTANCE: values of motion vectors of blocks are determined which blocks are adjacent with block where the motion vector should be determined. On the base of determined values of motion vectors of adjacent blocks, the range of search of motion vector for specified block is determined. Complexity of evaluation can be reduced significantly without making efficiency of compression lower.

EFFECT: reduced complexity of determination.

7 cl, 2 dwg

FIELD: compensation of movement in video encoding, namely, method for encoding coefficients of interpolation filters used for restoring pixel values of image in video encoders and video decoders with compensated movement.

SUBSTANCE: in video decoder system for encoding a video series, containing a series of video frames, each one of which has a matrix of pixel values, interpolation filter is determined to restore pixel values during decoding. System encodes interpolation filter coefficients differentially relatively to given base filter, to produce a set of difference values. Because coefficients of base filter are known to both encoder and decoder and may be statistically acceptably close to real filters, used in video series, decoder may restore pixel values on basis of a set of difference values.

EFFECT: efficient encoding of values of coefficients of adaptive interpolation filters and ensured resistance to errors of bit stream of encoded data.

5 cl, 17 dwg

FIELD: video encoding, in particular, methods and devices for ensuring improved encoding and/or prediction methods related to various types of video data.

SUBSTANCE: the method is claimed for usage during encoding of video data in video encoder, containing realization of solution for predicting space/time movement vector for at least one direct mode macro-block in B-image, and signaling of information of space/time movement vector prediction solution for at least one direct mode macro-block in the header, which includes header information for a set of macro-blocks in B-image, where signaling of aforementioned information of space/time movement vector prediction solution in the header transfers a space/time movement vector prediction solution into video decoder for at least one direct mode macro-block in B-image.

EFFECT: creation of improved encoding method, which is capable of supporting newest models and usage modes of bi-directional predictable (B) images in a series of video data with usage of spatial prediction or time distance.

2 cl, 17 dwg

FIELD: movement estimation, in particular, estimation of movement on block basis in video image compression application.

SUBSTANCE: method and device are claimed for conducting search for movement in video encoder system using movement vectors which represent difference between coordinates of macro-block of data in current frame of video data and coordinates of corresponding macro-block of data in standard frame of video data. A set of movement vector prediction parameters is received, where movement vector prediction parameters represent approximations of possible movement vectors for current macro-block, movement vector search pattern is determined and search is conducted around each movement vector prediction parameter from the set of movement vector prediction parameters using search pattern, and on basis of search result, the final movement vector is determined.

EFFECT: increased efficiency of video signals compression.

3 cl, 7 dwg

FIELD: physics.

SUBSTANCE: said utility invention relates to video encoders and, in particular, to the use of adaptive weighing of reference images in video encoders. A video encoder and a method of video signal data processing for an image block and the specific reference image index for predicting this image block are proposed, which use the adaptive weighing of reference images to increase the video signal compression, the encoder having a reference image weighting factor assigning module for the assignment of the weighting factor corresponding to the said specific reference image index.

EFFECT: increased efficiency of reference image predicting.

8 cl, 7 dwg

FIELD: physics, computing.

SUBSTANCE: invention relates to the field of coding and decoding of a moving image. In the method, at least one reference image for the processing of the field macroblock is selected from at least one reference image list, using information about reference image indexes, each at least one reference image selected is a field, and the parity of at least one reference field selected may be based on the parity of the field macroblock and the reference image index information.

EFFECT: efficient provision of information about reference image compensating motion, by reference image indexes determined in different ways, according to the coded macroblock modes.

10 cl, 12 dwg

FIELD: information systems.

SUBSTANCE: invention refers to video coders using adaptive weighing of master images. The video decoder for decoding data from video signal for the image, having multiple motion boxes, containing: the master image weighting coefficient module for accepting, at least, one master image index, thereat each one from the mentioned master image indexes is intended for independent indication not using any other indexes, one of the multiple master images, used for prediction of current motion box and weighting coefficient from the set of weighting coefficients for current mentioned one from mentioned multiple motion boxes.

EFFECT: increase of efficiency in predicting master images.

20 cl, 7 dwg

FIELD: information technology.

SUBSTANCE: method is offered to compress digital motion pictures or videosignals on the basis of superabundant basic transformation using modified algorithm of balance search. The algorithm of residual energy segmentation is used to receive an original assessment of high energy areas shape and location in the residual image. The algorithm of gradual removal is used to decrease the number of balance assessments during the process of balance search. The algorithm of residual energy segmentation and algorithm of gradual removal increase encoding speed to find a balanced basis from the previously defined dictionary of the superabundant basis. The three parameters of the balanced combination form an image element, which is defined by the dictionary index and the status of the basis selected, as well as scalar product of selected basic combination and the residual signal.

EFFECT: creation of simple, yet effective method and device to perform frame-accurate encoding of residual movement on the basis of superabundant basic transformation for video compressing.

10 cl, 15 dwg

FIELD: information technology.

SUBSTANCE: playback with variable speed is performed without picture quality deterioration. Controller 425 creates EP_map () with RAPI address in videoclip information file, dedicated information selection module 423 RAPI, image PTS with internal encoding, which is immediately preceded by RAPI, one of final positions of the picture with internal encoding, as well as the second, the third and the fourth reference pictures, which are preceded by the picture with internal encoding. The controller saves EP_map () in output server 426, i.e. controller 425 copies the value, close to given number of sectors (quantity of sectors, which can be read at one time during encoding process) of final positions for the four reference pictures (1stRef_picture, 2ndRef_picture, 3rdRef_picture and 4thRef_picture) to N-th_Ref_picture_copy, defines value of index_minus 1 on the basis of N-th_Ref_picture_copy and records it to disc.

EFFECT: effective process performance with constant data reading time.

8 cl, 68 dwg

FIELD: information technology.

SUBSTANCE: invention proposed contains videodecoder (200) and corresponding methods of videosignal data processing for image block with two reference frames' indices to predict this image block. The methods use latent scaling of reference images to improve video compressing. The decoder (200) contains latent scaling coefficient module (280) of reference images, which are used to determine a scaling coefficient value, corresponding to each of the reference image indices. Decoding operations contain receiving reference image indices with data, which corresponds to image block, calculation of latent scaling coefficient in response to image block location relative to reference images, indicated by each index of reference image, extraction of reference image for each of the indices, motion compensation relative to extracted reference image and multiplication of reference images, relative to which the motion compensation was performed, to a corresponding scaling value.

EFFECT: increase of decoding efficiency.

25 cl, 6 dwg

Up!