Compression of image with utilization of discontinuous cosine transformation of adaptively determined size of block based on dispersion

FIELD: engineering of circuit for compressing image signals, using blocks and sub-blocks of adaptively determined sizes of given coefficients of discontinuous cosine transformation.

SUBSTANCE: block size-setting element in encoder selects a block or sub-block of processed input pixels block. Selection is based on dispersion of pixel values. Blocks with dispersions greater than threshold are divided, while blocks with dispersions lesser then threshold are not divided. Transformer element transforms pixel values of selected blocks to frequency range. Values in frequency range may then be quantized, transformed to serial form and encoded with alternating length during preparation for transmission.

EFFECT: improved computing efficiency of image signals compression stages without loss of video signals quality levels.

4 cl, 5 dwg

 

I. Scope of the invention

The present invention relates to image processing. More specifically, the present invention relates to a compression scheme for image signals, using blocks and subblocks adaptive certain sizes of the encoded data of the coefficients of the discrete cosine transform.

II. Prior art

In the field of transmission and reception of video signals, such that are used for projection "movies" or "movies", made various improvements to the methods of image compression. Many of the common and the proposed systems using digital encoding. Digital coding ensures the reliability of communication lines, which resists distortion in the communication line, such as multi-path fading and intentional interference or noise signal, each of which would otherwise severely degrade image quality. In addition, digital methods facilitate the use of encryption of signals that are useful or even necessary for the government and many newly developed commercial applications broadcast.

High definition video is an area that takes advantage of advanced methods of image compression. the hen was first proposed transmission over the air signals high-resolution (or even passing through wires or fiber optic cables) seemed impractical because of the excessive demands for bandwidth. Typical wireless or other transmission system, as designed, was not easy to adapt with sufficient bandwidth. However, it was realized that the compression of digital video signals can be performed to a level that allows transmission using appropriate frequency bands. Such levels of compression signal associated with digital signal transmission, will allow the system to transmit with a lower power level with greater resistance to distortion in the channel, at the same time, occupying more desirable and usable bandwidth.

One compression method that is able to offer significant levels of compression and at the same time preserving the desired level of quality for video signals, uses blocks and subblocks adaptive certain sizes of the encoded data of the coefficients of the discrete cosine transform (DCT). This method will be referred to as the differential method adaptive cosine transform block size (DCARB). This method is disclosed in U.S. patent No. 5021891, entitled "Method and system for adaptive image compression block size", the right which the owner of the present invention and is incorporated into this description by reference. Ways DCT is also disclosed in U.S. patent No. 5107345, entitled "Method and system for adaptive image compression times the EPA unit", the right which the owner of the present invention and is incorporated into this description by reference. In addition, the method DCARB in combination with the method of differential transformation Quad-tree is disclosed in U.S. patent No. 5452104, entitled "Method and system for adaptive image compression block size", the right to which is also transferred to the owner of the present invention and is incorporated into this description by reference. The systems disclosed in these patents use the so-called "intraframe coding, where each frame of the image data is encoded, not paying attention to the content of any other frame. Using the method DCARB, the achievable data rate can be reduced from approximately 1.5 billion bits per second to approximately 50 million bits per second without noticeable degradation of image quality.

The way DCARB can be used to compress either black and white or color image, or a signal representing the image. Color input signal may be in the YIQ format (brightness/color), where Y is the sample brightness or luminance, I and Q samples are color or colors for each block of 4x4 pixels. They may also use other well-known formats, such as formats YUV and RGB (red-green-blue). Because of the low PR the spatial sensitivity of the eye to color most of the studies have shown, what components of the color subsampling with a factor of four in horizontal and vertical directions is acceptable. Thus, the signal can be represented by four components of the luminance and two chrominance components.

Using DCARB, the video signal is typically segmented into blocks of pixels for processing. For each block constituting the brightness and color are transferred in a block interleaver. For example, block 16×16 (pixel) may be filed in a block interleaver, which arranges or organizes a sampling of the image within each block of 16×16 in order to create blocks and composite sub-blocks of data for analysis of discrete cosine transform (DCT). The DCT operation is one way of converting the quantized time signal in the frequency representation of the same signal. When converting in frequency representation, it was shown that the methods DCT enable very high degrees of compression, since the unit of quantization can be designed to take advantage of the characteristics of the frequency distribution of the image. In a preferred implementation of one DCT 16×16 is applied to the first arrangement, four DCT 8×8 are applied to the second ordering, 16 DCT 4×4 apply for third have an order the deposits and 64 DCT 2× 2 are applied to the fourth ordering.

The DCT operation reduces spatial redundancy inherent in the source video. After DCT is performed, most of the energy of the signal tends to concentrate in the short DCT coefficients. Additional conversion differential conversion Quad-tree (DPDC) can be used to reduce the redundancy among the DCT coefficients.

For a block of 16×16 and each sub-block of DCT coefficients and the value DPDK (if used DPDK) are analyzed in order to determine the number of bits required to encode the block or sub-block. Then block or combination of sub-blocks, which requires the smallest number of bits for encoding is chosen to represent the segment of the image. For example, two sub-blocks 8×8, six sub-blocks 4×4 and eight sub-blocks 2×2 can be selected in order to represent the segment of the image.

The selected unit or combination of sub-blocks is then appropriately arranged in sequence in block 16×16. The values of the coefficients DCT/DPDK can then be subjected to frequency weighting a quantization and coding (such as variable length coding) in preparation for the transfer.

Despite the fact that the way DCARB described above,runs extremely well, it requires high computational cost. Therefore, a compact implementation and technical support of way can be difficult. Desirable alternative, which would do the technical support program more effective. A method and system for image compression, which are computationally more efficient, provided by the present invention in a particular way, as described below.

Summary of invention

The present invention is a system and method of image compression that use blocks and subblocks, adaptive certain data sizes of the coefficients of the discrete cosine transform. In one implementation of block 16×16 pixel is entered into the encoder. The encoder contains an element of the destination block size, which segments entered the unit pixel for processing. The purpose of the block size based on the dispersions of the input block and the subdivided blocks. Usually the area with large dispersions will be subdivided into smaller blocks, whereas areas with lower variance will not be subdivided, provided that the average size of block and sub-blocks come in various pre-defined ranges. Therefore, the first threshold dispersion unit is modified from its nominal value, depending on its average value, asetem the variance of the block is compared with a threshold and, if the variance is greater than the threshold, then the block is divided.

Prescribing information block size is supplied to the conversion element, which converts the pixel data into frequency domain data. The conversion is performed only on block and sub-blocks selected by assignment of the block size. These transformations are then subjected to quantization and transformation in serial form. For example, zig-zag scan may be used to convert data in serial form to create a data flow. The data stream can then be encoded using a variable length encoder in preparation for transmission. The encoded data is sent through the transmission channel to the decoder, where the pixel data is restored in preparation for display.

Brief description of drawings

The features, objectives and advantages of the present invention will become clearer from the detailed description below taken in conjunction with the drawings in which the same reference characters, respectively, are indicated on all drawings on which:

figure 1 - block diagram of the processing system image, which contains the system and method of appointment of the block size on the basis of the dispersion of the present invention;

figure 2 - sequence of operations, Illus ryuuma processing steps, included in the assignment block size on the basis of the dispersion;

figa, fig.3b and figs illustrate an example of a destination block size corresponding to the decomposition of the Quad-tree, and the corresponding data PQR.

Detailed description of preferred implementations

In order to facilitate the digital transmission of digital signals and to use the respective advantages, it is usually necessary to use some kind of compression of the signal. In order to achieve high resolution in the final image, it is also important to maintain high image quality. In addition, the computational efficiency is desirable for compact implementation and technical support, which is important in many applications.

The present invention provides a system or apparatus and method of image compression, which take into account both the image quality and computational efficiency when performing image compression. The image compression of the present invention is based on the discrete cosine transform (DCT). Usually the image to be processed in the digital domain, consists of pixel data divided into non-overlapping blocks of size N×N. two-Dimensional DCT can be performed on each block. Two-dimensional DCT is defined with the aid of the d following dependencies:

whereand

x(m, n) is the pixel location (m,n) in block N×M, and

X(k, l) is the corresponding DCT coefficient.

Since pixel values are not negative, component X(0,0) is always positive and usually has the greatest energy. In fact, for a typical image most of the energy conversion is concentrated around component X(0,0). This compactness property of the energy, respectively, makes the DCT method so attractive method of compression.

The compression method of the image of the present invention uses adaptive contrast coding in order to achieve an additional reduction of the bit rate. It is noticed that most natural images are composed of even a relatively slowly changing areas and rich areas, such as object boundaries and texture high contrast. Schema contrasting adaptive coding use this factor when assigning more bits to saturated areas and fewer bits for the less saturated areas.

Contrasting adaptive coding is also available to reduce the effect of partitioning. In the implementation of other ways of encoding DCT effect separation of the blocks, perhaps one is the most important harmful effect on image quality. In addition, the effect of partitioning tends to be more visible in bright areas of the image. However, it was found that the effect of partitioning decreases when DCT is used to specify a smaller size. The effect of partitioning is virtually invisible when using DCT 2×2, despite the fact that efficiency suffers in the calculation of bits per pixel. Therefore, contrasting adaptive coding can reduce the effect of partitioning by assigning smaller blocks (and thus, more bits) saturated areas and the large size of the blocks relatively to the blank areas.

Another characteristic of the present invention is that it uses intra-frame coding (spatial processing) instead of interframe coding (spatial-temporal processing). One of the reasons for selecting intra-frame coding is the high complexity of the receiver is required in order to process the signals interframe coding. Interframe coding requires an integral multiple frame buffers in addition to more complex processing schemes. In many applications, reduced complexity required for modern implementations.

The second reason for using intraframe coding is the fact that there may be situation or software material, which makes the scheme of the space-time coding to refuse or not be performed correctly. For example, movies are 24 frames per second may fall into this category, since link time (integration) because of a mechanical shutter is relatively short. A short integration time allows a higher degree of temporal overlap. The assumption of frame-by-frame correlation is disrupted for quick movement as it becomes jerky.

An additional reason for using intraframe coding is that the scheme of the space-time coding is more difficult to standardize, when the frequency of 50 Hz and 60 Hz line power. TV currently sends signals to either 50 Hz or 60 Hz. Using intra-frame scheme, which is a digital approach that can adapt to both 50 Hz and 60 Hz, or even to the movies 24 frames per second with the help of a compromise between frame rate relative to the spatial resolution.

For the purposes of image processing, the DCT operation is performed on the data of the pixels, which are divided into an array of non-overlapping blocks. Note that, although in the present description discusses the dimensions of the units as is equal to N×N, it can use the different sizes of blocks. For example, you can use a block size of N×M, where N and M are integers, and M is either greater or less than N. Another important aspect is that the block is divided into at least one level of sub-blocks, such as N/i×N/i N/i×N/j, N/i×M/j, etc. where i and j are integers. In addition, the approximate size of the block, as discussed in the present description, is a block of 16×16 pixels with the corresponding block and sub-blocks of DCT coefficients. Additionally, it seems that can be used in various other whole, such as both odd or even integer value, for example, 9×9.

Now referring to figure 1, depicted system 100 of the image processing, which contains the compression system of the present invention. System 100 of the image processing includes an encoder 102, which compresses the received video signal. The compressed signal is transmitted through the channel 104 of the transmission and received by the decoder 106. The decoder 106 decodes the received signal in the sample image, which can then be displayed.

Typically, the image is divided into blocks of pixels for processing. The color signal can be converted from space red-green-blue (RGB) space YC1C2where Y is the component of the brightness or luminance, a C1and C2are components of the chrominance or color. Because of the low spatial the Oh sensitivity of the eye to color many systems advanced quantum components of C 1and C2by a factor of four in horizontal and vertical directions. However, additional quantization is optional. The image with full resolution, known as format 4:4:4 can be either very useful or necessary in some applications, such as image, referred to as covering "digital cinema". Two possible views YC1C2are: view YIQ and YUV representation, both of which are well known in the art. It is also possible to use the YUV representation, known as YCbCr.

In the preferred embodiment, each of the components Y, Cb, and Cr is processed without podkashivaya. Then input unit 16×16 pixels is supplied to the encoder 102. The encoder 102 contains an element 108 of the destination block size, which performs the assignment block size in preparation for the compression of video. Element 108 of the destination block size defines the block decomposition of the block 16×16 on the basis of the perceived characteristics of the image in the block. The purpose of the block size divides each block of 16×16 into smaller blocks by way of the Quad-tree depending on the activity in block 16×16. Element 108 of the destination block size generates data Quad-tree, called PQR data, the length of which may be between the 1 and 21 bits. Therefore, if you assign block size determines that the block 16×16 should be disaggregated bit is set R PQR data, followed by four additional bits of data R corresponding to the four divided blocks 8×8. If you assign block size determines that any of the blocks 8×8 must be subdivided, then added four additional Q bits of data for each divided block of 8×8.

Now referring to figure 2, given the sequence of operations depicting the details of element 108 of the destination block size. The algorithm uses the variance of the block as a parameter when the decision unit block. Beginning at step 202, block 16×16 pixel is read. In step 204 calculates the variance v16 block 16×16. The variance is calculated as follows:

where N=16, a xi,j- the pixel in the i-th row, j-th column in the block N×N. At step 206, first change the threshold T16 dispersion to provide a new threshold T′16, if the average block size is between two predetermined values, then the variance of the block is compared with the new threshold T′16.

If the variance v16 is not greater than the threshold T16, then at step 208 writes the starting address of the block 16×16 bit R data PQR is set to 0 for the CSOs, to indicate that the block 16×16 is not divided. The algorithm then reads the next block of 16×16 pixels. If the variance v16 is greater than the threshold T16, then in step 210 bit R data PQR is set to 1 to indicate that the block 16×16 should be subdivided into four blocks of 8×8.

Four blocks 8×8, i=1:4, are discussed in turn for additional units, as depicted in step 212. For each block of 8×8 calculates the variance v8iat step 214. In step 216, first change the threshold T8 dispersion to provide a new threshold T′8, if the average block size is between two predetermined values, then the variance of the block is compared with this threshold.

If the variance v8inot more than the threshold T8, then at step 218 writes the starting address of the block of 8×8, and the corresponding bit Q, Qiis set to 0. Then processed the next block of 8×8. If the variance v8igreater than the threshold T8, then at step 220 the corresponding bit Q, Qiset to 1 to indicate that the block of 8×8 must be subdivided into four blocks of 4×4.

Four blocks 4×4, ji=1:4, are discussed in turn for additional units, as shown in step 222. For each block of 4×4 calculates the variance v4ijnusage 224. In step 226, first change the threshold T4 dispersion to provide a new threshold T′4, if the average block size is between two predetermined values, then the variance of the block is compared with this threshold.

If the variance v4ijnot more than the threshold T4, then in step 228 is written to the start address of a block of 4×4, and the corresponding bit R, Pijis set to 0. Then processed the next block of 4×4. If the variance v4ijgreater than the threshold T4, then at step 230 the corresponding bit R, Pijset to 1 to indicate that the block of 4×4 must be subdivided into four blocks 2×2. In addition, records addresses of 4 units 2×2.

Thresholds T16, T8 and T4 can be pre-defined constants. This is known as a tough decision. Alternative can be implemented adaptive and flexible solution. Flexible solution changes the thresholds for dispersions depending on the average value of the pixel blocks of 2N×2N, where N can be 8, 4 or 2. Consequently, the functions of averages of pixels can be used as thresholds.

To illustrate consider the following example. Let predefined thresholds dispersion for component Y is equal to 50, 1100 and 880 for blocks 16×16, 8×8 and 4×4, respectively. In other words, T16=50, T8=1100 and T16=880. Let d is Amazon averages 80 and 100. Assume that the calculated variance for a block of 16×16 60. As 60 and its average value 90 more than T16, block 16×16 is divided into four sub-blocks 8×8. Assume that the calculated dispersion for blocks 8×8 equal 1180, 935, 980 and 1210. As two of the blocks 8×8 have a variance that exceed T8, these two blocks are additionally classified for up to eight sub-blocks 4×4. Finally, suppose that the variance of the eight blocks of 4×4 equal 620, 630, 670, 610, 590, 525, 930 and 690 with the corresponding averages 90, 120, 110, 115. Since the average value of the first block 4×4 falls within the range (80, 100), its threshold will be reduced to T4=200, which is less than the 880. Thus, this unit is 4×4 is also divided into seven blocks of 4×4. The resulting assignment of the block size shown on figa. The corresponding decomposition tree quadrants depicted in fig.3b. In addition, PQR data generated by this assignment block size shown on figs.

Note that a similar procedure is used to assign block dimensions for the components of C1and C2color. The constituent colors can be thinned horizontally, vertically or in both directions.

In addition, note that although the purpose of the block size described as a top-down approach, in which the beginning is estimated the largest block (16× 16 in this example), may be used instead of the bottom up approach. The bottom up approach is first to estimate the smallest blocks (2×2 in this example).

With reference to figure 1 describes the remainder of the system 100 of the image processing. Data PQR together with the addresses of the selected blocks are served in the element 110 DCT. Element 110 DCT uses data PQR for performing discrete cosine transform of the appropriate size relative to the selected blocks. Only the selected blocks need to be subjected to the DCT processing.

System 100 image processing can selectively contain DPDK 112 to reduce redundancy among the regular DCT coefficients. Constant coefficient occurs in the upper left corner of each DCT block. The constant coefficients are usually large compared with variable coefficients. The difference in size makes it difficult to design effective variable length encoder. Thus, it is advantageous to reduce the redundancy among the permanent factors.

Element 112 DPDK performs two-dimensional DCT relatively constant coefficients, taking one 2×2. Since blocks 2×2 in blocks of 4×4, two-dimensional DCT is performed with respect to four of constant coefficients. This DCT 2×2 is called the differential transform tree quadran is s, or DPDC, four constant coefficients. Then constant coefficient DPDK along with three neighboring constant coefficients unit 8×8 are used to calculate DPDC the next level. Finally, the constant coefficients of the four blocks 8×8 block 16×16 are used to calculate DPDK. Therefore, in block 16×16 there is one valid constant coefficient, and the rest are variable coefficients corresponding to the DCT and DPDCH.

Conversion factors (as DCT and DPDK) served in the device 114 quantization quantization. In a preferred embodiment, the DCT coefficients quanthouse using frequency weighting masks (CVM) and scale factor quantization. CVM is a table of the frequency scales of the same dimensions as the block of the input DCT coefficients. Frequency weights apply different weights to different DCT coefficients. The weights are intended to highlight the input samples having the frequency content, to which the human visual system is more sensitive, and to suppress the samples having the frequency content, to which the visual system is less sensitive. Weight can also be designed on the basis of factors such as viewing distance and so on

Weight is selected n is the basis of empirical data. The method for constructing the weighing masks for coefficients of 8x8 DCT disclosed in ISO/IEC JTC1 CD 109918 "Digital compression and coding of half-tone still images - part 1: Requirements and guidelines". International standards organization, 1994, which is incorporated into this description by reference. Usually constructed from two CVM, one component for luminance and one for chrominance components. Table CWM for sizes 2×2, 4×4 blocks obtained by thinning, and 16×16 - using interpolation tables for a block of 8×8. The scale factor controls the quality and bit rate of the quantized coefficients.

Therefore, each DCT coefficient quantized in accordance with the relationship:

where DCT(i, j) - input DCT coefficient, fwm(i, j) is the frequency weighting mask, q is the scale factor, and q(i, j) is quantized coefficient. Note that depending on the sign of the DCT coefficient of the first member inside the brackets is rounded up or down. The coefficients DPDK also quanthouse using appropriate weighting masks. However, you may have many tables masks and applied to each of the components Y, Cb and Cr.

The quantized coefficients are fed into the Converter 116 in serial form zigzagoon the CSOs scan. The Converter 116 serialized scans blocks of quantized coefficients zigzag manner in order to create a serialized stream of quantized coefficients. The number of different scan templates and templates other than zigzag, can also be selected. The preferred method uses a size 8×8 block to the zigzag scan, despite the fact that you can use other sizes.

Note that the Converter 116 in serial form can be located either before or after the device 114 quantization. The results are the same.

In any case, the stream of quantized coefficients is fed to the encoder 118 of variable length. The encoder 118 of variable length can be used coding with variable length of zeros, followed by encoding the encoding of hatmana. This method is discussed in detail in the aforementioned U.S. patent No. 5021891, 5107345 and 5452104 and summarized in the present description. Encoder with variable length selects the quantized coefficients and allocates zero from non-zero coefficients. Zero values are referred to as variable length and is encoded by way of hatmana. Non-zero values separately coded way of Chapman.

Modified coding Afma is and quantized coefficients is also possible and is used in the preferred implementation. Here after zigzag scanning encoder variable length to define a pair of variable length/size of each block of 8×8. These pairs of variable length/size is then encoded by way of Chapman.

Codes Chapman are constructed from either the measured or theoretical statistics of the image. It was observed that most natural images are composed of empty or relatively slowly changing areas and rich areas, such as object boundaries and high-contrast texture. Encoders of hatmana with frequency-spatial transformation, such as DCT, use these features when assigning more bits saturated areas and fewer bits blank areas. Usually the encoders of hatmana use look-up tables for encoding variable length and non-zero values. Usually multiple tables, and in the present invention preferred 3 tables, despite the fact that can be used 1 or 2 when needed.

The compressed image signal generated by the encoder 102, and transmitted to the decoder 106 through the channel 104 of the transmission. Data PQR, which contain information of the destination block size, also available at the decoder 106. The decoder 106 includes a decoder 120 of variable length, which decade is the duty to regulate variable length value and a nonzero value.

The output signal of the decoder 120 of variable length is served in the reverse Converter of the successive forms of the zigzag scan, which arranges the coefficients in accordance with the scheme of scanning. Inverter 122 of the successive forms zigzag scan takes PQR data in order to participate in an appropriate ordering of the coefficients in the composite block of coefficients.

The composite block is fed into the device 124 inverse quantizer for inverse processing due to the use of frequency weighting masks.

The block of coefficients is fed to the element 126 TPDC (reverse DPDK), followed by the element 128 ADCP (inverse DCT), if it was applied to differential conversion of the Quad-tree. Otherwise, the block of coefficients is served directly in the element 128 ADCP. Element 126 TPDC and element 128 ADCP perform the inverse transform coefficients to generate a block of pixel data. The pixel data can be then be interpolated, converted in the form of red-green-blue, and then stored for future display.

Accordingly, the system and method for image compression that fulfill the purpose of the block size on the basis of dispersion of pixels. The purpose of the size of the Loka based dispersion has several advantages. As the discrete cosine transform is performed after a defined block size allows efficient calculation. Intensive computationally, a conversion must be performed only for the selected blocks. In addition, the process of unit selection is effective, because the variance of the values of the pixels is mathematically easy to calculate. Another advantage of the destination block size on the basis of dispersion is that it is based on perception. The variance of the pixels is the criterion of activity in the unit and provides an indication of the presence of contours, textures, and so He seeks to collect details of the block is much better than criteria, such as average values of the pixels. Therefore, the scheme based on the dispersion, the present invention assigns a smaller window areas with large footprints, and large blocks of more than smooth areas. In the best quality can be achieved in the restored image.

Another important advantage is that, since the purpose of the block size is performed before quantization, greater flexibility is provided when the control bit rate and quality. Since the threshold dispersion is adapted to the local average, small blocks are assigned even in relatively dark areas of the I. This keeps the details in all areas that are barely visible above the visibility threshold. In addition, the image compression on the basis of dispersion provides a graceful degradation of the image quality, when the scale factor quantization varies from small to large quantities, in contrast to methods such as MPEG compression of moving images, proposed by a group of experts in the field of moving images). This is particularly critical for applications such as digital cinema.

In wide demand for digital video piracy is a serious threat. Insert digital watermarking is an important requirement in order to prevent copyright infringement and loss of income. So as inserting watermarking is performed in the image areas that are important to understand the purpose of the block size on the basis of the dispersion is a natural candidate to insert a "watermark".

The previous description of the preferred implementations are provided to enable any person skilled in the art to make or use the present invention. Various modifications to these implementations will be easily understood by professionals in this area of technology, and the basic principles defined in the present description can be applied to others the shM implementation options without the use of inventive ability. Therefore, it is not intended that the present invention is limited to the implementation described in the present description, and must comply with the General framework that is compatible with the principles and new features, disclosed in the present description.

1. The method of determining the destination of the block size for the input data block of pixels to be used when compressing mentioned input unit that reads the data block of pixels, generate the assignment of block size on the basis of the dispersions of the values of the pixels of the above-mentioned data block of pixels and the subdivided blocks of the above-mentioned data block of pixels, and the said step of generating includes the steps, which determine the variance of the values of the pixels for the above-mentioned block pixel data, comparing said variance with a preset threshold, and the said threshold is a function of the average value of values of pixels of the evaluated unit, decide on the division of the above-mentioned block, if the mentioned dispersion more of the above-mentioned threshold, if the decision is in the unit mentioned block, then repeat the steps of determining, comparing and making decisions for each of the subdivided unit to the until will not be satisfied with a predefined criterion, denoted by the try, as mentioned destination block size, each block, which in the future should not be subdivided, and builds a data structure containing information about the mentioned purpose of the block size.

2. The method according to claim 1 in which the said threshold varies for each unit level.

3. The method according to claim 1, in which referred to a pre-defined criterion that no longer repeat the steps of determining, comparing and making decisions based on pre-selected minimum data block size of pixels.

4. The method according to p. 1, in which said dispersion is determined in accordance with the following equation:

where N is the dimension of the block;

xi,j-the pixel in the i-th row, j-th column in the block of NxN.

5. System image compression to compress the data block of pixels containing means to the destination block size to select the above-mentioned unit or subdivided blocks above the block to be compressed, based on the dispersions of the values of the pixels of the above-mentioned data block of pixels and the subdivided blocks of the above-mentioned data block of pixels, and the above-mentioned means of the destination block size is configured to determine a variance value of the pixels for the above-mentioned block pixel data, comparing said variance with what anee a given threshold, moreover, the above-mentioned threshold is a function of the average value of values of pixels of the evaluated unit, the decision unit of the above-mentioned block, if the variance is more mentioned predetermined threshold, in this case, if the decision is in the unit mentioned block, then repeating step of determining the variance, comparison with a preset threshold and a decision unit for each of the subdivided unit to the until will not be satisfied with a predefined criterion, and denote, as mentioned destination block size, each block, which in the future should not be subdivided conversion tool for converting pixel data mentioned the selected unit or subdivided blocks of data in the frequency domain, quantization means for quantizing the mentioned data to frequency domain conversion tool in serial form to the mentioned scanning quantized data in a serialized data stream, and a means of encoding with variable-length encoding mentioned serialized data stream in preparation for the transfer.

6.The system according to claim 5, in which the said threshold varies for each level of the subsection the expression.

7. The system according to claim 5, in which referred to a pre-defined criterion to no longer be distinguished, based on pre-selected minimum achievable size of the data block of pixels.

8. The system according to claim 5, in which the above-mentioned conversion tool performs the discrete cosine transform.

9. The system according to claim 5, in which the above-mentioned conversion tool performs the discrete cosine transform, followed by the differential transformation of the Quad-tree.

10. The system according to claim 5, in which the above-mentioned conversion tool in serial form contains the device zigzag scan.

11. The system of claim 10, in which the said device zigzag scan uses the size of the 8x8 block to the zigzag scan.

12. The system according to claim 5, in which the said means of encoding with variable-length contains the Huffman encoder.

13. The system under item 12, which mentions the Huffman encoder uses a lot of viewing tables to encode values are variable-length and non-zero value.

14. The system under item 13, in which there are three lookup tables.

15. The system according to claim 5, in which said dispersion is determined in accordance with the following equation:

the de N - the dimension of the block;

xi,j-the pixel in the i-th row, j-th column in the block of NxN.

16. The compression method of the data block of pixels in the image, namely, that reading a block of pixel data, generate the assignment of block size on the basis of the variance value of the pixels of the above-mentioned data block of pixels and the subdivided blocks of the above-mentioned data block of pixels, and the said step of generating further comprises the steps, which determine the variance of the values of the pixels for the above-mentioned block pixel data, comparing said variance with a preset threshold, and the said threshold is a function of the average value of values of pixels of the evaluated unit, decide on the division of the above-mentioned block, if the variance more of the above-mentioned threshold, if the mentioned decision on the subdivision mentioned block, then repeat the steps of determining, comparing and making decisions for each of the subdivided unit to the until will not be satisfied with a predefined criterion, and designate, as the destination block size, each block, which are not further subdivided, forming a data structure containing information relative to the said destination block size transform said data pixels of the selected block, as the criminal code is re-mentioned data structure, in view of the frequency domain, quantuum mentioned data in the frequency domain based on a human-perceivable characteristics of the image, scan mentioned quantized data in a serialized data stream, and encode mentioned a serialized stream of data in preparation for transmission.

17. The method according to clause 16, in which the said threshold varies for each unit level.

18. The method according to clause 16, which referred to a predefined criterion, in order not to repeat the steps of determining, comparing and making decisions based on pre-selected minimum data block size of pixels.

19. The method according to clause 16, which referred to the discrete cosine transform performed during this phase of the conversion.

20. The method according to clause 16, in which the discrete cosine transform is performed with subsequent differential conversion of a Quad-tree in the transformation stage.

21. The method according to clause 16, in which the zigzag scanning is performed during the above-mentioned scanning stage.

22. The method according to item 21, in which the aforementioned zigzag scanning is performed by using the block size is 8×8.

23. The method according to clause 16, in which the coding Huffman perform during upon the wrapped phase encoding.

24. The method according to item 23, in which the said Huffman coding uses a lot of viewing tables to encode values are variable-length and non-zero value.

25. The method according to paragraph 24, in which there are three lookup tables.

26. The method according to clause 16, in which said dispersion is determined in accordance with the following equation:

where N is the dimension of the block;

xi,j-the pixel in the i-th row, j-th column in the block of NxN.

27. The compression system data block of pixels in the image that contains the tool for reading the read data block of pixels, generating tool for generating a destination block size based on the dispersions of the values of the pixels of the above-mentioned data block of pixels and the subdivided blocks of the above-mentioned data block of pixels, and the said means of generating includes means for determining for determining a variance value of the pixels for the mentioned blocks of pixel data, a comparator for comparing the above-mentioned dispersions with a pre-defined threshold, and the said threshold is a function of the average value of values of pixels of the evaluated unit, means a decision for a decision on the subdivision mentioned block, if the variance more of the above-mentioned threshold, if the aforementioned means of decision-making is intended for decision-making unit of the above-mentioned block, then define, compare, and decide for each of the subdivided unit to the until will not be satisfied with a predefined criterion,and means to indicate, as mentioned destination block size, each block, which are not further subdivided.

28. The system according to item 27 in which the said threshold varies for each unit level.

29. The system according to item 27, which referred to a pre-defined criterion based on pre-selected minimum data block size of pixels.

30. The system according to item 27, which mentions the conversion tool uses the discrete cosine transform.

31. The system according to item 27, which mentions the conversion tool uses the discrete cosine transform, followed by the differential transformation of the Quad-tree.

32. The system according to item 27, which mentions the scan tool uses zig-zag scan.

33. The system according to item 27 in which the said means of encoding uses Huffman coding.

34. System p. 33, in which the said Huffman coding uses a lot of viewing tables to encode values are variable-length and non-zero values.



 

Same patents:

FIELD: engineering of devices for transforming packet stream of information signals.

SUBSTANCE: information signals represent information, positioned in separate, serial packets of digital format data. These are transformed to stream of information signals with time stamps. After setting of time stamps, which are related to time of arrival of data packet, time stamps of several data packets are grouped as packet of time stamps, wherein, in accordance to realization variant, size of time stamps packet equals size of data block.

EFFECT: improved addition of data about time stamps to data packets with fixed size.

6 cl, 29 dwg

FIELD: methods and devices for memorization and processing of information containing video images following one another.

SUBSTANCE: from each image recorded prior to current time appropriately at least one image area is selected and aperture video information is recorded with placement information. from video-information at least one mixed image is generated with consideration of appropriate placement information. Mixed image is utilized for display in accordance to movement estimation, movement compensation or error masking technology frames.

EFFECT: decreased memory resource requirements for memorization of multiple previously received images.

3 cl, 4 dwg

FIELD: multimedia technologies.

SUBSTANCE: method includes at least following stages: determining, whether current value of processed coefficient of discontinuous cosine transformation is equal or less than appropriate threshold value, used in current time for quantizing coefficients of discontinuous cosine transformation of image blocks of common intermediate format, and if that is so, then value of discontinuous cosine transformation coefficient is set equal to zero, then currently used threshold value is increased for use as threshold value with next processing of discontinuous cosine transformation coefficient, in opposite case currently used threshold value is restored to given original threshold value, which is used as threshold value for next processing of discontinuous cosine transformation coefficient; determining, whether increased threshold value is greater than given upper limit of threshold value, and if that is so, than increased threshold value is replaced with given upper limit.

EFFECT: higher quality.

8 cl, 4 dwg

FIELD: protection of video information against unauthorized copying.

SUBSTANCE: proposed method using watermarks to protect video information against unauthorized copying by changing scale of pattern in the course of copying includes introduction of watermark in original video signal with different scales. Watermark is maintained in each scale for preset time interval sufficient to enable detector circuit in digital-format video recorder to detect, extract, and process information contained in watermark. Watermark scale is changed by end of preset time interval preferably on pseudorandom basis to ensure appearance of each of all scales in predetermined scale variation range as many times as determined in advance. In this way definite scale possessing ability of watermark recovery to initial position and size can be identified and used for watermark detection.

EFFECT: enhanced reliability, facilitated procedure.

24 cl, 7 dwg

FIELD: converting code of received video sequence using extrapolated movement data received from video sequence.

SUBSTANCE: proposed method for code conversion involves reception of first bit stream of compressed picture data having some coding parameters. These parameters may relate to GF-structure of picture frames, picture frame size, to parameter showing if frames presented in input bit stream are picture fields or frames, and/or if they form picture frames presented in bit stream, direct or interlaced sequence. First and second movement vectors are obtained from input bit stream and used together with weighting coefficients to extrapolate third movement vector for output bit stream of compressed picture data. Output bit stream that differs from input one by one or more parameters is outputted as converted-code output signal.

EFFECT: provision for minimizing of or dispensing with movement estimation in code conversion process.

22 cl, 4 dwg, 1 tbl

FIELD: coding elementary-wave data by means of null tree.

SUBSTANCE: proposed method includes generation of elementary-wave ratios pointing to image. In the process bits of each elementary-wave ratio are associated with different bit places so that each place is associated with one of bits of each elementary-wave ratio and associated bits are coded with respect to each place of bits to point to null tree roots. Each place of bits is also associated only with one of bits of each elementary-wave ratio. Computer system 100 for coding elementary-wave ratios by means of null tree has processor 112 and memory 118 saving program that enables processor 112 to generate elementary-wave ratios pointing to image. Processor 112 functions to code bits of each place to point to null tree roots associated with place of bits.

EFFECT: enhanced data compression speed.

18 cl, 7 dwg

The invention relates to the field of shorthand, more specifically to the introduction and detection of watermark data in the data stream, such as, for example, video stream

The invention relates to the field of steganography, in particular for detecting a digital watermark

The invention relates to image compression to reduce bandwidth requirements of digital video decoder

Camcorder // 2191483
The invention relates to video equipment, in particular, to the design professional camcorder related to the type of camcorders, the peculiarity of which is the use of image compression and digital recording in the memory the received audio and video data and storing them in memory in a compressed form

FIELD: coding elementary-wave data by means of null tree.

SUBSTANCE: proposed method includes generation of elementary-wave ratios pointing to image. In the process bits of each elementary-wave ratio are associated with different bit places so that each place is associated with one of bits of each elementary-wave ratio and associated bits are coded with respect to each place of bits to point to null tree roots. Each place of bits is also associated only with one of bits of each elementary-wave ratio. Computer system 100 for coding elementary-wave ratios by means of null tree has processor 112 and memory 118 saving program that enables processor 112 to generate elementary-wave ratios pointing to image. Processor 112 functions to code bits of each place to point to null tree roots associated with place of bits.

EFFECT: enhanced data compression speed.

18 cl, 7 dwg

FIELD: converting code of received video sequence using extrapolated movement data received from video sequence.

SUBSTANCE: proposed method for code conversion involves reception of first bit stream of compressed picture data having some coding parameters. These parameters may relate to GF-structure of picture frames, picture frame size, to parameter showing if frames presented in input bit stream are picture fields or frames, and/or if they form picture frames presented in bit stream, direct or interlaced sequence. First and second movement vectors are obtained from input bit stream and used together with weighting coefficients to extrapolate third movement vector for output bit stream of compressed picture data. Output bit stream that differs from input one by one or more parameters is outputted as converted-code output signal.

EFFECT: provision for minimizing of or dispensing with movement estimation in code conversion process.

22 cl, 4 dwg, 1 tbl

FIELD: protection of video information against unauthorized copying.

SUBSTANCE: proposed method using watermarks to protect video information against unauthorized copying by changing scale of pattern in the course of copying includes introduction of watermark in original video signal with different scales. Watermark is maintained in each scale for preset time interval sufficient to enable detector circuit in digital-format video recorder to detect, extract, and process information contained in watermark. Watermark scale is changed by end of preset time interval preferably on pseudorandom basis to ensure appearance of each of all scales in predetermined scale variation range as many times as determined in advance. In this way definite scale possessing ability of watermark recovery to initial position and size can be identified and used for watermark detection.

EFFECT: enhanced reliability, facilitated procedure.

24 cl, 7 dwg

FIELD: multimedia technologies.

SUBSTANCE: method includes at least following stages: determining, whether current value of processed coefficient of discontinuous cosine transformation is equal or less than appropriate threshold value, used in current time for quantizing coefficients of discontinuous cosine transformation of image blocks of common intermediate format, and if that is so, then value of discontinuous cosine transformation coefficient is set equal to zero, then currently used threshold value is increased for use as threshold value with next processing of discontinuous cosine transformation coefficient, in opposite case currently used threshold value is restored to given original threshold value, which is used as threshold value for next processing of discontinuous cosine transformation coefficient; determining, whether increased threshold value is greater than given upper limit of threshold value, and if that is so, than increased threshold value is replaced with given upper limit.

EFFECT: higher quality.

8 cl, 4 dwg

FIELD: methods and devices for memorization and processing of information containing video images following one another.

SUBSTANCE: from each image recorded prior to current time appropriately at least one image area is selected and aperture video information is recorded with placement information. from video-information at least one mixed image is generated with consideration of appropriate placement information. Mixed image is utilized for display in accordance to movement estimation, movement compensation or error masking technology frames.

EFFECT: decreased memory resource requirements for memorization of multiple previously received images.

3 cl, 4 dwg

FIELD: engineering of devices for transforming packet stream of information signals.

SUBSTANCE: information signals represent information, positioned in separate, serial packets of digital format data. These are transformed to stream of information signals with time stamps. After setting of time stamps, which are related to time of arrival of data packet, time stamps of several data packets are grouped as packet of time stamps, wherein, in accordance to realization variant, size of time stamps packet equals size of data block.

EFFECT: improved addition of data about time stamps to data packets with fixed size.

6 cl, 29 dwg

FIELD: engineering of circuit for compressing image signals, using blocks and sub-blocks of adaptively determined sizes of given coefficients of discontinuous cosine transformation.

SUBSTANCE: block size-setting element in encoder selects a block or sub-block of processed input pixels block. Selection is based on dispersion of pixel values. Blocks with dispersions greater than threshold are divided, while blocks with dispersions lesser then threshold are not divided. Transformer element transforms pixel values of selected blocks to frequency range. Values in frequency range may then be quantized, transformed to serial form and encoded with alternating length during preparation for transmission.

EFFECT: improved computing efficiency of image signals compression stages without loss of video signals quality levels.

4 cl, 5 dwg

FIELD: technology for compacting and unpacking video data.

SUBSTANCE: for each pixel of matrix priority value is determined, pixels difference value is calculated, priority values utilized for calculating value of pixels priority are combined in one pixels group, pixel groups are sorted, pixel groups are saved and/or transferred in accordance to their priority in priority matrix, while aforementioned operations are constantly repeated, while values of pixel group priorities are repeatedly determined anew, priority matrix for any given time contains pixel groups sorted by current priorities, and also preferably stored first and transferred are pixel groups, which have highest priority and still were not transferred.

EFFECT: simple and flexible synchronization at different transfer speeds, width of transfer band, resolution capacity and display size, respectively.

2 cl, 8 dwg, 1 tbl

FIELD: engineering of systems for encoding moving images, namely, methods for encoding moving images, directed at increasing efficiency of encoding with use of time-wise remote supporting frames.

SUBSTANCE: method includes receiving index of supporting frame, standing for supporting frame, pointed at by other block, providing movement vector for determining movement vector of current block, and determining movement vector of current block with utilization of supporting frame index, denoting a supporting frame.

EFFECT: increased efficiency of encoding in direct prediction mode, decreased number of information bits for frame, in which scene change occurs.

3 cl, 6 dwg

FIELD: engineering of systems for encoding moving image, namely - methods for encoding moving image, directed at increase of encoding efficiency with use of time-wise remote supporting frames.

SUBSTANCE: in the method in process of encoding/decoding of each block of B-frame in direct prediction mode movement vectors are determined, using movement vector of shifted block in given frame, utilized for encoding/decoding B-frame, and, if type of given frame is time-wise remote supporting frame, one of movement vectors, subject to determining, is taken equal to movement vector of shifted block, while another one of movement vectors, subject to determining, is taken equal to 0.

EFFECT: increased encoding efficiency in direct prediction mode, decreased amount of information bits for frame, wherein a change of scene occurs.

2 cl, 6 dwg

Up!