Method and device for frame-accurate encoding of residual movement based on superabundant basic transformation to increase video image condensation

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

 

The technical field to which the invention relates.

The present invention relates to the field of seals and, in particular, to methods and devices for sealing the video.

The level of technology

The sequence of images may take a large amount of memory space and require a very large bandwidth transmission at the performance in non-digital form. Point-to-point digital video has become feasible a few years ago following the progress of computer networks and technology seals signals.

The attempt of standardization for digital videoplotter was taken around 1988. Currently, the Committee Expert group on cinematography (TFE) (MPEG) with ISO/IEC (international organization for standardization/ international electrotechnical Commission) has completed the standards as MPEG-1 and MPEG-2; MPEG-4 is also completed, but still accept new proposals. In addition, the CCITT (consultative Committee for international Telegraph and telephony) has developed a set of recommendations - N, n and n+, which focus on applications with low bit rates. All these attempts use in the standardization of a two-stage procedure for sealing the sequence. The first stage uses a motion estimation algorithm to compensate for the emission to create a predicted frame for the current frame using the previous frame, this calculates the difference between the current frame and the predicted frame, which is called residual image motion (OID) (MRP). The second step in the standard procedure consists in encoding OID using discrete cosine transform (DCT) (DCT). These are based on the DCT system does not have good performance in all circumstances. At low bit speeds necessary for personal video based on DCT system cause noticeable distortion and visible block artifacts. For applications with high visual quality, such as DVD to be produced by compacting factor can be quite low.

The residual image motion can be encoded using other transform-based methods. For example, you can also use the discrete wavelet transformation (dwt) (DWT) and surpanaka basic conversion. Zakhor and Neff presented in U.S. patent No. 5699121 coding system residual motion based algorithm surpanaka basic transformations, called the search for reconciliation. It was first proposed by Mallat and Zhang in IEEE Transaction in Signal Processing, vol.41, No.12, Dec. 1993. The video encoder Zakhor and Neff and improves the visual quality and s/n power (PNSR) compared to the standard based on DCT video encoders. However, their system is actimagine and characterization seal is not optimized due to adapted to the specific case designs for an agreed basic coding position and quantization of transform coefficients. Therefore, there is a need for a new method of coding on the basis of surpanaka conversion, who can provide both speed and efficiency.

This information is from the prior art provided for the purpose, as suggested by the applicant, to make known information on the features essential to the present invention. Not made mandatory assumptions and should not be construed that any one of the preceding information is the prototype for the present invention.

The invention

The purpose of the present invention is to provide a method and a device frame encoding residual motion on the basis surpanaka basic transformations to videoplotter. In accordance with the object of the present invention proposes a method of encoding a residual image using basis functions from serpanos library containing the following steps: receive a residual image that has the size and energy; and carry out the decomposition mentioned residual image in the list of one or more elements, each of which represents a basic function of serpanos library, and mentioned step decomposition mentioned residual image includes the following steps: (i) identify the region of samadani the residual image to represent the item using the segmentation algorithm residual energy; (ii) create a subset of basis functions from serpanos library, with each basis function in this subset is consistent with the scope of substitution in a pre-specified limits; (iii) identify the element within the subset of basis functions, and the above mentioned element is used to represent the area of the substitution and the above-mentioned element has parameters; (iv) quantum mentioned element and modify the parameters of the element in a form suitable for encoding; (v) encode mentioned quantized element, subtract the above item from the area of the substitution in the residual image, thereby lowering the energy of the residual image, and use based on the tree quadrants element the encoder, in order to reduce the size of the residual image; and (vi) comparing the reduced size of the residual image or the reduced energy of the residual image with predetermined criteria, and repeat steps (i) through (vi) until such time as these pre-defined criteria will not be achieved; through this code referred to the residual image and reduce its size to a pre-specified level.

In accordance with another object of the present invention a device for encoding a residual image using basis functions from serpanos library, comprising: means for receiving the residual is about image, moreover, the above image has the size and energy; and means for decomposition mentioned residual image in the list of one or more elements, each of which represents a basic function of serpanos library, and the said means for decomposing mentioned residual image includes: (i) a means to identify the area of the substitution in the residual image to represent the item using the segmentation algorithm residual energy; (ii) means for creating a subset of basis functions from serpanos library, with each basis function in this subset is consistent with the scope of substitution in a pre-specified limits; (iii) means for identify the element within the subset of basis functions, and the above mentioned element is used to represent the area of the substitution and the above-mentioned element has parameters; (iv) means for quantizing the mentioned element and to modify the parameters of the element in a form suitable for encoding; (v) a means for encoding mentioned quantized element, subtracting the above item from the area of the substitution in the residual image, which lowers the energy of the residual image, and to use based on the tree quadrants of the coder element to reduce the size of the residual image; and (vi) redtwo for comparing the reduced size of the residual image or the reduced energy of the residual image with predetermined criteria; whereby encode mentioned residual image and reduce its size to a pre-specified level.

In accordance with another object of the present invention features a computer program product containing a machine-readable medium recorded thereon a computer program for executing the method of encoding the residual image using basis functions from serpanos library containing the following steps: receive a residual image that has the size and energy; and carry out the decomposition mentioned residual image in the list of one or more elements, each of which represents a basic function of serpanos library, and mentioned step decomposition mentioned residual image includes the following steps: (i) identify the area of the substitution in the residual image to represent the item using the segmentation algorithm residual energy; (ii) create a subset of basis functions from serpanos library, with each basis function in this subset is consistent with the scope of substitution in a pre-specified limits; (iii) identify the element within the subset of basis functions, and the above mentioned element is used to represent the area of the substitution and the above-mentioned element has parameters; (iv) quantuum at manucy element and modify the parameters of the element in the form, suitable for encoding; (v) encode mentioned quantized element, subtract the above item from the area of the substitution in the residual image, thereby lowering the energy of the residual image, and use based on the tree quadrants of the element encoder in order to reduce the size of the residual image; and (vi) comparing the reduced size of the residual image or the reduced energy of the residual image with predetermined criteria, and repeat steps (i) through (vi) until such time as these pre-defined criteria will not be achieved; through this code referred to the residual image and reduce its size to a predetermined level.

Brief description of drawings

Figure 1 illustrates the General scheme of systems videoplotter that use sorpano basic conversion, and related ways of encoding according to one variant of implementation of the present invention.

Figure 2 is an example of a residual image motion, processed by one embodiment of the present invention.

Figure 3 illustrates a simple dictionary with 16 databases for use with one embodiment of the present invention.

Figure 4 describes the entire process of cell decomposition based surpanaka basis according to one variant of implementation of this breath is retene.

Figure 5 describes the step, performed by the segmentation algorithm residual energy (ASOA) (RESA) according to one variant of implementation of the present invention.

6 illustrates the first step of ASOA according to one variant of implementation of the present invention.

7 illustrates the second step of ASAE: the scheme of the growth of horizontally according to one variant of implementation of the present invention.

Fig illustrates the second step of ASAE: the scheme of the growth of vertically according to one variant of implementation of the present invention.

Figure 9 describes the search item perform matching using the algorithm gradual removal (APA) (PEA) according to one variant of implementation of the present invention.

Figure 10 illustrates how to create podlogar basis of negotiation and candidates we seek a position according to one variant of implementation of the present invention.

11 illustrates the rapid calculation of the energy field according to one variant of implementation of the present invention.

Fig illustrates the parameters for the same element according to one variant of implementation of the present invention.

Fig is an example of a map element according to one variant of implementation of the present invention.

Fig is a block diagram of the algorithm, illustrating the process by which tiravanija element, according to one variant of implementation of the present invention.

Fig is a block diagram of the algorithm, illustrating the decoding of a compressed residual signal according to one variant of implementation of the present invention.

Detailed description of the invention

The present invention represents a new encoder for encoding a residual image on the basis of surpanaka transformations used for systems videoplotter with compensated movement. This invention similar to the previous encoders search agreeing that they decompose the residual image in the list of elements that represents the basic functions of surpanaka dictionary. The process of finding items, however, is performed using the segmentation algorithm residual energy (ACOE) and algorithm gradual removal (ASD). Basic dictionary can be very large in order to characterize the signs, often appearing in the residual image motion. To find the element, ACOE identifies the approximate shape and position of areas with high energy in the residual image motion, so that it may be found good agreement by comparing with a smaller subset of bases in the dictionary. Further, ASD sequentially removes the images of candidates from consideration by the seat reservation calculate the energy of the search window, whereby reduced the computation time needed to find the best coordination. When agreed the item is found, the residual image is updated by removing part characterized by this element. The foregoing steps for finding items and update the residual images are repeated until then, until you have achieved the desired bit rate or quality.

The invention introduces a new modular scheme of quantization to search for harmonization in serpanos the basis that changes the procedure of finding items. The coefficients obtained directly from the conversion are continuous floating-point values that you want to quantize for optimal digital encoding of bit resources. In the search algorithm negotiation, you must use a loop quantizer, where each found item is first quantized and then used to update the residual image. As such, each element affects the choice of subsequent elements. If the quantizer is specified before start coding, as in the previous search methods of coordination, it is difficult to optimize the quantization scheme, since the optimal design of the quantizer depends on the statistics of the selected list element modules. CX is mA quantization according to the present invention selects the adaptive quantizer in the search process elements.

In addition to the elemental module based on serpanos converting the encoder, you must pass a pointer of the selected basis and the position of the elements. The invention includes a method to efficiently encode information element position. The distribution of element positions forms a two-dimensional map, where the pixel values of one or zero represent, respectively, the presence or absence in each position. The method is similar to the tree quadrants provides coding card provisions. Information modules and the base pointer is included in the coding position. Elements for different color channels video (Y, U, V) are encoded independently.

The element parameters are transmitted after they are encoded in a compacted version of the residual image. For the decoding process, the decoder recovers the residual image through the interpretation of the encoded bit stream back into the element parameters and the combination of information elements for the formation of the recovered stream of residual images, which are then combined with the compensated image motion, in order to form the restored stream.

The present invention is a method of encoding a residual image motion, containing the following steps: FD is mirouet elemental decomposition of the residual image in space surpanaka basis using a modified search algorithm negotiation; choose modular quantizer; encode map of the provisions of the module, and the pointer for the selected basis. The present invention further proposes a method of decoding a residual signal that is encoded using the above encoding method.

Figure 1 illustrates the related processing performed by the device 10 videoplotter, which uses the encoder 20 of the residual image, according to one variant of implementation of the present invention. The video frame is first processed by the block 30 motion estimation, which compares the current frame with one or two keyframes. In most cases, the objects in the video image change their position in successive frames, while the background remains the same. Because keyframes are transmitted to the video decoder 12, some areas in the reference frame can be used to construct the current frame, the block 30 motion estimation determines the region in the reference frames, which are similar to the regions in the current frame. The compensator 32 motion produces the difference between these similar areas and combines them as a residual image motion. The ratio of provisions between similar regions are represented as motion vectors, which are processed by the encoder 34 of the motion vector. First, the residual image is agenie processing unit 40 elemental decomposition, and then the element encoder 42 condenses the resulting elements. The coded motion vectors and the elements are combined into a single bit stream multiplexer 22. Sealed video image is transmitted to and stored by device 24, which can deliver video in a compressed format to the decoder 12.

The lower part of figure 1 illustrates the decoder 12, the demultiplexer 26 separates the multiplexed video signal, sending appropriate bits to the decoder 36 motion vector and the decoder 28 residual image, respectively. Block 38 recovery movement forms the frame of the prediction from the reference frame and motion vectors. The decoder 28 residual image restores the residual image. These two signals, namely the frame prediction and the residual frame are summed to generate the final reconstructed frame.

Figure 2 represents the estimated residual image motion for color channel Y. the Original residual image has both negative and positive values. For the purpose of displaying the residual image as an image with 256 gray levels of the pixel values of the residual image are shifted and scaled so that the pure gray color means zero, while the black and white colors represent, respectively, the denier is haunted and positive values. For example, the residual image contains multiple high-energy regions, which correspond to the moving objects in the video image.

Most of the methods of compaction signals, convert the raw data into some more compact format by means of mathematical transformations of various kinds. Some mathematical transformation, such as DCT and fiberboard, using the full basis, which forms the matrix is invertible transformation. Recently, considerable attention was attracted corpany basis and the associated transformation algorithms. The number of bases in the dictionary surpanaka basis much more than the dimensionality of the original data. The advantage surpanaka basis is that the transformed coefficients are more effective in the presentation of a valid indication in the original signal. There are many mathematical methods to build a basic vocabulary for different signals. For residual video motion built several dictionaries and proved that they are well covered signs in the residual images. For example, the basic dictionary-based separable Gabor functions described Neff and Zakhor in "Very Low Bit Rate Video Coding Based on Matching Pursuits", IEEE Transactions on Circuits and Systems for Video Technology, Feb. 1997, 158-171, and basic dictionary-based Haar functions described Vleeschouwer and Maq New dictionaries for matching pursuit video coding", Proc. of the 1998 International Conference on Image Processing, vol.1, 764-768. Figure 3 is a simple approximate dictionary containing 16 bases. Any of the above dictionaries can be used with the present invention. In particular, in relation to the above dictionary of Gabor features 400 two-dimensional explicitly mentioned functions. However, in reality it involves much more basic structures implicitly, because each of these 400 two-dimensional functions can be placed in any possible position in the image. Using a frame size of 176×144 pixel assumes that the dictionary actually contains 400×176×144=5.7 million base structures, which makes it eminently corpany. Converting directly using the "search algorithm negotiation described S.Mallat and Z.Zhang in "Matching Pursuits With Time-Frequency Dictionaries", IEEE Transaction in Signal Processing, vol.41, No.12, Dec. 1993, would require an extremely large number of calculations to determine the conversion factors. Search agreeing to seal video, invented Zakhor and Neff in U.S. patent No. 5699121, reduces the burden of calculation, however, remains expensive computationally. The present invention offers a way of converting a residual image on the basis of a common vocabulary, which is performed by the block 40 elemental decomposition and path encoding conversions the bathrooms coefficients, which is a task element of the encoder 42.

Work unit 40 elemental decomposition is fully described in figure 4 according to one variant of implementation. The first step (block 61), the executable unit 40 elemental decomposition is to find the initial search area. This step is implemented by the segmentation algorithm residual energy (ACOE), and one alternative implementation is shown in figure 5. ASOA based on the idea of common building areas. First, he selects the block 2×2 as a starting point for building area (block 70). This step requires the separation of the residual image into blocks 16×16, as shown in Fig.6. The energy that is the sum of the squares of all of the pixel intensities is calculated for each block, and the block with the highest energy is determined, for example, as block 71 shown in Fig.6. The block 71 is further divided into four sub-blocks 8×8, and determines the sub-block 72 with the highest energy. In this subsection 72 size 8×8 is also defined by subsection 73 size 2×2 with higher energy, and this block will be used as a starting point for building up the field.

The next step of ASAE (block 74 shown in figure 5)consists in verifying unit 2×2 to the left of the current scope. Fig.7 illustrates this step of ASOA. The threshold is calculated dynamic range is and how

T=AE*max(7-AU, 5)/10,

where AU is the number of blocks that are added to the left of the starting block, and AE is the average energy per block 2×2 the current scope. If the energy is a proven unit 2×2 more than the current threshold, a checked block 2×2 grouped with the current region, forming together a new large current region. Otherwise, on this side found the ending point, and we grouped the blocks together. Similar symmetric verifies that the block 2×2 to the right of the current scope. Continue to build first the left side then the right side until, until we find the end points on both sides or until the width reaches 32 (whatever came first). After this step forms a horizontal band rectangle 75, and the dimension of the tape is 2*2m, 1≤m≤16.

The last step of ASAE (block 76 figure 5) is the capacity region vertically on the basis of step 75, as shown in Fig. Suppose that the width of the tape equal to 75 W. Consider the tape rectangle 2*W over current scope with threshold:

T=AEs*max(7-AUs, 5)/10,

where AUs is the number of rectangles 2*W, which added over the starting tape, a AEs is the average energy of the rectangle 2*W, included in the current scope. If proven rectangle 2*W is the energy that is greater than the threshold, include it in the current scope. Otherwise, on this side found the endpoint. Similar symmetric verifies that the unit 2*W below the current scope. Continue to build up first from the top and bottom until then, until we find the end points on both sides or until the height reaches 32 (whatever came first). In the end it will be a rectangle 77, which has the dimension 2n*2m, 1≤n,m≤16.

Again referring to figure 4, illustrates the process of finding the nearest consistent basis from a given dictionary (block 62). The degree of concordance between the basis and the residual image is represented by an absolute value (module), their scalar product, which is called cell module, when this great module implies a good agreement. The process of determining this module requires the calculation of several scalar products and select the one with the highest module as the current item. This process can be the slowest part of the algorithm of the search agreement. In a classical search algorithm negotiation would need to compute the scalar product between the residual image and each of the millions of items in the dictionary to determine the module. For example, in the prototype unit 16*16 with the highest energy in the residual image is selected as the start of the region search: each basic structure centered at each position in the selected block, and calculates the scalar product between the base structure and the corresponding residual area. For a dictionary with 400 bases this process requires 256×400=102400 computing scalar products. Fig.9 illustrates the new process of searching for approval in accordance with the present invention.

The resulting rectangle 77 ASOA on Fig provides initial evaluation form high-energy characteristic. It is used to filter the bases in the dictionary that have the form, too different from the rectangle ASOA. Then create a subset of bases matching (block 80). Assume that the width and height of the rectangle is equal to w and h respectively, then formed podlogar containing all the bases with forms, defined respectively by the width (width) and height (height), which satisfy the conditions:

w-tw1≤width≤w+tw2 and h-th1≤height≤h+th2,

where tw1, trw2, th1 and th2 are values set to limit the size of the basis. These values can be changed and adjusted according to the structure of the dictionary. The largest and smallest sizes of scanned bases are illustrated as rectangles 90 and 91 shown in figure 10. For example, the block 80 is a simple example under dictionary, containing four basis.

ASOA can further assess the situation of high-energy features in the residual image. The provisions of the candidate bases for negotiation are selected around the center right is Galenika 77 ASOE (block 81). Figure 10 shows a small rectangle 92, the center of which is the same as that of the rectangle 77 ASOA. It is assumed that all pixels in the rectangle 92 will operate as a centre for proven residual area. Rectangle 94 figure 10 is an example, the center of which is the point 93, or the upper left corner of the rectangle 92. It is assumed that the width (ws) and height (hs) of the rectangle 92 must be variables in relation to the rectangle 77 ASOA. This ratio is as follows:

ws=2*min(w/2+1,6) and hs=2*min(h/2+1,6).

The size of the rectangle 92 may be other rules or simply be captured in the embodiment. The idea of bases is that a good agreement is located around the center of the rectangle 77 ASOA. Further, any provisions in the rectangle 92, which already contain the center of the element, will not be considered as any new items. Point 95 figure 10 is an example. It should be noted that the prototype does not put such restrictions. The idea of this type of restriction is that if one element ensures good contact, you should remove the energy around its center without adding too much excess energy on the boundary. Essentially, the search algorithm negotiation is not desirable to return to the same position to get the second element. This limitation of the ban on repeating what their position is almost no effect on the encoding performance and can make simpler the encoding status information of the element.

The next processing step (block 89 figure 9) is called the algorithm gradual removal (APA) to search for remaining approval. It does not depend on the method used to generate the dictionary of the test basis and test set of provisions. For example, the UPA will work if podlogar is a whole dictionary, and a set of provisions candidate is a set of coordinates that contain the entire image. ASD is a way of finding the nearest basis negotiate more effectively by the gradual removal of candidates comparison from consideration. This contrasts with the classic search agreement, which compares all the bases-candidates in all possible positions. First, the maximum module is set to zero (block 82). Considered the following basis b(k,l) (block 83), where k and l represent the width and height of a two-dimensional basis functions. Formed area of the same size, centered on the same position candidate r(k,l,p) in the residual image (block 84). Unit 85 compares ||r(k,l,p)||, the energy of r(k,l,p) with maximum current module (Mm)to decide whether there is a need to compute the scalar product between r(k,l,p) and b(k,l). To explain this operation, remember mathematical inequalities triangles:

|<r(k,l,p),b(k,l)>|≤||r(k,l,p)|| ||b(k,l)||.

The goal of the search for the harmonization of the Oia is to find the maximum |< r(k,l,p), b(k,l)>|. Suppose that the current maximum modulus equal to Mm. If the basis b(k,l) at position R corresponding to the remainder r(k,l,p) satises |r(k,l,p)|| ||b(k,l)||≤Mm, then:

|<r(k,l,p), b(k,l)>|≤||r(k,l,p)|| ||b(k,l)||≤Mm.

In this case, you do not need to compute the scalar product |<r(k,l,p), b(k,l)>|, and the region r(k,l,p) moves to the next position. The norm of the basis of ||b(k,l)|| can be calculated in advance (in fact most of the bases normalized, namely, ||b(k,l)||=1, and the only overkill for this test then consists in calculating the energy of r(k,l,p). An efficient algorithm to determine ||r(k,l,p)|| is described below.

Suppose that there are n different dimensions base height {v1v3,..., vn} and m different dimensions basic width {h1h2,..., hm}that are sorted by increasing. The dimensionality of the search rectangle is equal to hs*ws, and the upper left point of the search rectangle is equal to p(x,y). Energy values hs*ws*n*m can be calculated through the following four steps:

Step 1: Calculate the energy for s=hm+k columns (11 shows an example of the columns). These columns are centered in (x-hm/2+i,I), where i=0, 1,..., s-1. Their height is equal to v1. Their energy is represented as1,0(0), C1,1(0),..., C1,s(0) and is calculated as:

C1,i(0)=e(x-hm/2+i, u-v1/2)+...+e(x-hm/2+i, y)+...+(x-h m/2+i, y+v1/2),

where e(x,y) represents the energy of the pixel in position (x,y).

Energy for the following s columns with the same coordinates as in the above strips, and the length of v2can be calculated as:

With2,i(0)=C1,i(0)+Extra(v2-v1energy pixels, i=1, 2,...s.

In the General case we have:

Withj,i(0)=Cj-1,i(0)+Extra(vj-v(j-1)energy pixels, i=1, 2,..., s; j=1, 2,...n.

Step 2: Calculate the energy of columns that are shifted vertically from the column in step 1, using:

Cj,i(a)=Cj,i(a-1)-e(x-hm/2+i, u-v1/2+a-1)+e(x-hm/2+i, y+v1/2+a), a=1..., hs,

where a represents the number of vertical panning appropriate.

Step 3: Calculate energy areas with height vj, (j=1,...,n) and width h1h2,..., hmand center (x, y+a), v=0, 1,..., hs), using:

Sj,i(0,a)-Cj,(hm-h1)/2(a)+...-+Cj,hm/2(a)+...+Cj,(hm+h1)/2(a)

Sj,2(0,a)=Sj,i(0,a)+Extra(h2-h1energy columns.

In the General case:

Sj,i(0,a)=Sj,i-1(0,a)+Extra(h2-h(i-1)energy columns, i=1,..., m.

Step 4: Calculate the energy of the first set of fields with a base length of vertical vj(j=1,..., n) and the base length of the horizontal hi(i=1,..., m) and the center (x+b, y+a), (b=1,...ws and a=1,...,hs) with:

Sj,i(b,a)=Sj,i(b-1, a)-Cj,(hm-hi)/2+b-1(a)+Cj,(hm+hi)/2+b(a).

The maximum is Odul can consistently be updated in the process of seeking approval. this may gradually to determine the search space. Some bases may have the same dimensions, thus one calculates the energy can replace the computation of several scalar products. The performance of the APU also refers to how often is good agreement (not necessarily the best match). Due to the fact that large areas always contain more energy, the bases of larger dimension are checked first.

If ||r(k,l,p)||>Mm, the block 86 is executed to calculate the scalar product (R) between r(k,l,p) and b(k,l). Block 87 compares the absolute value of R with the current maximum Mm. If |R|>Mm, as |R| is set to the new Mm and records the pointer and the position basis. Regardless of this, we continue back to block 84 to until all provisions of the search will not be checked. Then the blocks 83-88 run again until, until you have checked all the bases candidates. Finally, it turns out the item, which includes three parameters: 1. The pointer basis in the dictionary, which gives the best agreement; 2. The location of the best matching in the residual image with coordinates (x, y); and 3. The scalar product (R) between the basis and the residual image. Fig shows an example of an item on the residual image.

With the ova referring to figure 4, step after finding the element is in the recording element parameters (block 63). Note at this stage that there is no quantization cell module is not running. Block 64 decision will decide when we get to the quantization of the element. His work is problem-dependent speed control, a certain system videoplotter. If the compacting factor is fixed, the block 64 will check if still bits for more items. Because he hasn't done any actual coding, you should evaluate the bits used to encode the current elements. Let the "Bip" is the average number of bits for encoding basic pointers and regulations, "Bm(i)" represents the valid number of bits for the i-th element of a module without quantization. When allocating one bit for the sign of the scalar product (p) the number of used bits for n elements is estimated as:

Used bits = n*(Bip+1)+∑(Bm(1)+Bm(2)+...Bm(n)),

where "Bip" is initialized according to the experimental data for the first residual frame and is installed as a valid value for the last frame. Bm(i) may be known for several modules. It is important that the module will be quantified later and will give a smaller number to be using bits than estimated at the moment. Thus the om, at this stage will typically less items than what can be coded. If the system wants to achieve a certain quality, which is determined by the root mean square error (MSE) (MSE) encoded residual image compared to the actual residual image block 64 will compare the current achieved an RMSE of target RMS. RMS after the introduction of one element is updated according to the following equation:

CKO(n)=CKO(n-1)-p(n)*p(n),

where MSE(n) represents the MSE after using n elements and p(n) is the scalar product of the n-th element. First, RMS, or RMS (0)is the energy source of the residual image. After quantization is performed, MSE(n), is likely to increase, and therefore no longer achieved the target RMS. In the end, if the bits are available or the problem of quality is not achieved, the residual image will be updated on the basis of the current element (block 65), followed by the search for another item, again starting with the block 61. Otherwise, if the bit or qualitative goal is reached, marks the block 66 for the design of quantization. Update the residual image is one step for the standard search algorithm negotiation - can be mathematically described as:

r(k,l,p)=r(k,l,p)-p(n)*b(k,l).

All areas not covered by the current element, b is to be unchanged.

The project quantizer (block 66) based on the value of the minimum module (Minm), found at the moment. The step size of quantization (HQ) (QS) set as follows:

Any items found to this point, will be quantized using the above SHK in simple scalar quantization with a reading medium. Following this, the residual image is again updated according to now quantized list element module 67. Assume that the elemental ratio before and after quantization are equal, respectively, p(i) and q(i) (i=1,..., n). Assume that the corresponding bases are b(i), (i=1,..., n). The residual image after n aquantance elements is:

E(n)=(Initial residual)-p(1)b(1)-p(2)b(2)-...-p(n)b(n).

His energy ||E(n)|| also known. There are two ways to calculate the residual energy after quantization. The first way is to simply calculate the residual image after quantization as:

EQ(n)=(Initial residual)-q(1)b(1)-q(2)b(2)-...-q(n)b(n).

Another way is to recursively update. Assume that the quantization error for p(i) is equal to Δp(i). Then the residual image, in which only p(n) procentowe, features:

EQ(1)=E(n)-ΔR(n)b(n) and ||EQ(1)||=||E(n)||+Δp(n)*Δp(n)-2ΔR(n)<E(n),b(n)>.

The remainder with the quantization of p(n) and p(n-1) becomes:

EQ(2)=EQ(1)-Δp(n-1)g(n-1).

This is when the relation is true recursively and can be written as:

EQ(i)=EQ(i-1)-Δp(n-i+1)g(n-i+1), i=1, 2,...n, EQ(0)=E(n).

The corresponding energy is equal to:

||EQ(i)||=||EQ(i-1)||+Δp(n-i+1)Δp(n-i+1)-2*Δp(n-i+1)<EQ(i-1),g(n-i+1)>.

Finally, we take EQ(n) and ||E(n)||, which are the starting point for further found items. It is important that the elements can be in any order for the recursive update - the update does not need to be conducted in the order in which matching elements.

Because the modules of elements quanthouse, now will have more elements to achieve the goal of controlling the speed or quality. Therefore, the block 68 is executed to find additional items. The process is the same as the blocks 61-63. However, at this stage, the modules will be quantize immediately. We now need to deal with the elements, modules, which is less than (SHK-SHK/4), without dropping by setting their size quantization to zero. Used the scheme is given below.

1. If the module element is greater than (SHK-SHK/4), the quantizer uses SHK.

2. Otherwise, if the module element is greater than (HQ/2-SHK/8), it is quantized as the value SHK/2.

3. Otherwise, if the module element is greater than (HQ/4-SHK/16), it is quantized as the value SHK/4.

4. Otherwise, if the module element is greater than (SHK/8-HQ/32), it is quantized as the value SHK/8.

In practice, as a rule, enough TRICORONA down although you can use more levels.

After the block 68 is executed to the real logical unit speed control (block 69). Since all elements quanthouse at this stage in the cycle, we can estimate the achieved quality or the number of used bits. When achieving the goal of the seals, the system enters the element encoder 42. Otherwise, the residual image will be updated on the basis of the quantized element module, and the system returns to block 68, to find the next element. Color video image contains several channels, i.e. channels Y, U and V Block 40 separate items will be used for each channel independently. With this scheme, each channel can have its own bitmap resource or desirable quality. There are several ways to select the bits that can be used for bit allocation of resources to different channels.

All items are to elemental encoder 42 for removal in a compact form. The present invention examines the distribution of elements for each channel as a two-valued map, as shown in Fig. Black pixels represent the elements in their proper position, while white pixels are no elements in this position. To encode the provisions containing the elements you can use is to use a method similar to the Quad-tree, although, as it is easy to understand, you can use other methods. Other parameters of each element can be encoded after the information of the position of the elements, for example, using coding with variable length, however, other coding methods may be used, as known to experts in this field. Procedure coding for the signal parameters of the elements illustrated in Fig and described in more detail below.

The first step of the encoding elements consists in the decomposition of the entire map elements, for example, as shown in Fig, n*n blocks (block 101). The value of n can be either 16 (channel Y)or 8 (for channels U and V). For each block of n*n, if the unit has no elements, it displays zero bit; otherwise, it displays a single bit, and the block is processed further for placing elements in the decoder. Using the decomposition procedure of the Quad-tree, and it is summarized in the following four steps.

Step 1. To initialize the list element block (PUB) (LAB) with one element - the block of n*n.

Step 2. Take one element e of PUB. If the size of e is equal to 1*1, list all element parameters except for the provisions, namely, should be withdrawn underlying index, the modulus and the sign of the scalar product, then go to step 4; otherwise go to step 3.

Step 3. Output bits combin, is of the four subblocks of e: a 1and2and3and4where ai(i=1, 2, 3, 4) is a unit if the corresponding sub-block has the item, and zero otherwise. Place all subblocks i value andi1, at the end of the PUB and return to step 2.

Step 4. To check if it is empty PUB. If it is not empty, return to step 2. Otherwise, the encoding is completed for a single block of n*n.

Index basis and the module element can be encoded using a variable length coder to save bits, since these signal parameters can be evenly distributed. Information of the position of the element can be encoded implicitly by writing procedures decomposition data 0/1 bit. The encoding of variable length can be used to encode bits of the combination of the four sub-blocks: AAA. There are 15 species combinations for bits combinations of elements, AAA, and it should be noted that 0000 is impossible. However, some combinations, such as 1000, occur with much higher probability than the other combinations. The probabilities of the various combinations can be evaluated in experiments and used to create a draft table of variable length. Further, it should be noted that the probability distribution can be variable for different channels and different densities of the elements. So you can use mn is the number of tables, and information categories block may be coded first, so that the decoder knows which table should be used for the purposes of decoding.

Fig illustrates the element decoder 46, which performs an operation inverse to that performed element by the encoder 42. First element decoder 46 receives one bit representing the state for the current block of n*n. If the value is equal to the unit, it is processed according to the procedure of decomposition of a symmetric Quad-tree. First block of n*n is divided into four sub-blocks. Bits combinations of elements for these four sub-blocks are decoded using the inverse variable length coding (DPC) (VLC). Then all the sub-blocks with a value of 1 is placed in the list element block (PUB). The PUB is dynamically updated by a recursive decomposition of each element in the PUB and getting his bits combinations of elements. If an item from the PUB is a block 1*1, the pointer and the module elemental basis should be decoded using table opposite the cap should then read the bit representing the sign of the scalar product. the element decoder for a single block of n*n end if the PUB is empty.

The decoded signal element parameters can then flow to the residual reducing agent 48 that forms one channel of the residual image with the help of pic is BA classical search coordination. First, all pixels in the residual image are set to zero. Then, each element is added one by one using the following procedure. Let q(i) and b(i,k,l) represent, respectively, the coefficient of the i-th element and the corresponding basic two-dimensional matrix. If (x(i), y(i)) represents the location of the i-th element of the matrix q(i)*b(i,k,l) is added to a residual image, built before the position (x(i), y(i))to obtain the new current image. This process is repeated until, until all items are added to the channels. When each channel is decomposed, the process ends and the residual image is restored.

Those familiar with the previous technique of coding, based on the search for reconciliation, you will see several benefits associated with the methods according to the present invention. The decomposition of elements on the basis of surpanaka basic space accelerated through the procedure more accurate estimates of energy and due to the algorithm gradual removal candidates. The project quantizer element of the complete module is selected circuit element decomposition, whereas previous analogues was determined quantizer before started the conversion. Finally, the process of encoding elements more efficient, because rhythm is e decomposition based on Quad-tree apply the spatial relationship between the elements. In particular, the prototype collects all of the elements in a one-dimensional list, thereby making it difficult for their efficient encoding, compared with the present invention.

Embodiments of the invention described here, obviously, can vary in many ways. Such changes should not be regarded as a departure from the essence and scope of the invention, and all such modifications as would be obvious to a person skilled intended for inclusion in the scope the following claims.

1. The method of encoding the residual image using basis functions from serpanos library containing the following steps: a) obtain a residual image that has the size and energy; and b) carry out the decomposition mentioned residual image in the list of one or more elements, each of which represents a basic function of serpanos library, and mentioned step decomposition mentioned residual image includes the following steps: (i) identify the area of the substitution in the residual image to represent the item using the segmentation algorithm residual energy (ASOA); (ii) create a subset of basis functions from serpanos library, each basis function in this subset is consistent with the area of the substitution in advance satanicreds; (iii) identify the element within the subset of basis functions, and the above mentioned element is used to represent the area of the substitution and the above-mentioned element has parameters; (iv) quantum mentioned element and modify the parameters of the element in a form suitable for encoding; (v) encode mentioned quantized element, subtract the above item from the area of the substitution in the residual image, thereby lowering the energy of the residual image, and use based on the tree quadrants of the element encoder in order to reduce the size of the residual image; and (vi) comparing the reduced size of the residual image or the reduced energy of the residual image with predetermined criteria, and repeat steps (i) through (vi) until such time as these pre-defined criteria will not be achieved; through this code referred to the residual image and reduce its size to a pre-specified level.

2. The method according to claim 1, wherein the step of identifying the element within the subset of basis functions is performed using the algorithm gradual deletion that removes the basic functions of a subset of the basis functions by comparing the basic functions being evaluated at the moment, with the previously estimated basis function.

3. The method according to claim 1, wherein the step of identifying the area of the substitution includes generating direct is Opalenica ASOA.

4. The method according to claim 3, in which the step of identifying the area of the substitution contains the identification of the initial region in the residual image with the highest energy, and the increment from him rectangle, ASAE.

5. The method according to claim 1, wherein the step of identifying the element within the subset of basis functions contains the definition of the scalar product between the basis function and the area of the substitution, with the maximum absolute value of this scalar product indicates a better match.

6. The method according to claim 3, in which the rectangle of ASOA compared with the basic functions in serpanos the library, and the basic function, which is sufficiently consistent with the rectangle ACOA, is placed in the subset of basis functions.

7. The method according to claim 1, wherein the step of quantization element contains the definition of the quantizer on the basis of comparison between the item and the area of the substitution.

8. The method according to claim 1, wherein the predefined criterion is determined based on the desired size of the bitstream.

9. Device for encoding a residual image using basis functions from serpanos library containing: a) a means for obtaining the residual image, and the above image has the size and energy; and b) means for decomposing mentioned residual image in lane is very of one or more elements, each of which represents a basic function of serpanos library, and the said means for decomposing mentioned residual image includes: (i) a means to identify the area of the substitution in the residual image to represent the item using the segmentation algorithm residual energy; (ii) means for creating a subset of basis functions from serpanos library, with each basis function in this subset is consistent with the scope of substitution in a pre-specified limits; (iii) a means to identify the item within the subset of basis functions, and the above mentioned element is used to represent the area of the substitution and the above-mentioned element has parameters; (iv) means for quantizing the mentioned element and to modify the parameters of the element in a form suitable for encoding; (v) a means for encoding mentioned quantized element, subtracting the above item from the area of the substitution in the residual image, which lowers the energy of the residual image, and to use based on the tree quadrants of the coder element to reduce the size of the residual image; and (vi) means for comparing the reduced size of the residual image or the reduced energy of the residual image with predetermined criteria; through CEG the code mentioned residual image and reduce its size to a pre-specified level.

10. The computer-readable storage medium containing a sequence of operations, the execution of which on the computer provides the ability to perform a method of encoding a residual image using basis functions from serpanos library containing the following steps: a) obtain a residual image that has the size and energy; and b) carry out the decomposition mentioned residual image in the list of one or more elements, each of which represents a basic function of serpanos library, and mentioned step decomposition mentioned residual image includes the following steps: (i) identify the area of the substitution in the residual image to represent the item using the segmentation algorithm residual energy; (ii) create a subset of basis functions from serpanos library, with each basis function in this subset is consistent with the scope of substitution in a pre-specified limits; (iii) identify the element within the subset of basis functions, and the above mentioned element is used to represent the area of the substitution and the above-mentioned element has parameters; (iv) quantum mentioned element and modify the parameters of the element in a form suitable for encoding; (v) encode mentioned quantized element, subtract mentioned ELEH the UNT from the area of the substitution in the residual image, thereby reduce the energy of the residual image, and use based on the tree quadrants of the element encoder in order to reduce the size of the residual image; and (vi) comparing the reduced size of the residual image or the reduced energy of the residual image with predetermined criteria and repeating steps (i) through (vi) until such time as these pre-defined criteria will not be achieved; through this code referred to the residual image and reduce its size to a pre-specified level.



 

Same patents:

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: 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: 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: 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: 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: 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 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

The invention relates to the field of digital signal processing

The invention relates to the formation, transmission and processing of the television program guide for broadcast television service

FIELD: physics.

SUBSTANCE: invention is related to processing of video image and more specifically to methods and devices for assignment of motion vectors. Method is suggested for processing multiple motion vectors for video image, which is interpolated with application of the first reference shot and second reference shot. Every reference shot contains content chart and involves division of video image interpreted into multiple zones, determination of number of motion vectors passing through one of multiple zones on the basis of content charts of the first reference shot and second reference shot, and formation of motion vector for assignment to single zone on the basis of number of motion vectors that pass through one zone.

EFFECT: elimination of distortions of tailing and modularity; reduction of complexity of interpolation operations for process of motion vectors assignment for interpolated shots that contain overlapping and empty zones.

60 cl, 9 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: technological processes.

SUBSTANCE: invention is related to the field of coding and decoding of moving image. The method of reference image selection for field macroblock processing consists of the following: definition of at least one list of reference images by using order of imaging, at that at least one list of reference images has information about indices of reference images; selection of at least one reference image for processing of field macroblock, from at least one list of reference images, by using information about indices of reference images, moreover, at least one selected reference image is a field, and evenness of at least one selected reference field is based on evenness of current image field macroblock and information about index of reference image; at that, if information about index of reference image represents even number of index, then at least one selected reference field has the same evenness of the field, that the field macroblock, and if information about index of reference image represents uneven number of index, then at least one selected reference field has evenness of field that is different from evenness of field macroblock.

EFFECT: improves efficiency of moving image coding with application of multiple reference images.

5 cl, 12 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: 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: video processing.

SUBSTANCE: method for encoding includes determining if all flags of current layer included in the specified unit area are equal flags of the main layer, setting pre-defined flag of forecast according to the result of determining, and if it is determined that flags of current layer are equal to flags of the main layer, flags of current layer are omitted and flags of the main layer are inserted and said forecast flag into the bit stream.

EFFECT: improvement of encoding efficiency of various flags used in multilayer scaled video coder-decoder based on interlayer correlation; method and device are suggested for efficient encoding of various flags used in multilayer scaled video coder-decoder based on interlayer correlation.

21 cl, 12 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: engineering of systems for encoding digital video signals, in particular, indication of values of quantization parameters in video encoding system.

SUBSTANCE: method and device for encoding a digital video series are claimed, where indication of quantization parameter is given out in encoded bit stream for use during decoding. Indication of information concerning the quantization parameter is ensured by insertion of SQP value - series level quantization parameter value. In particular, instead of encoding absolute values of parameters of quantization of image/section, indication of difference ΔQP between series level quantization parameter SQP and QP of image/section, is given out.

EFFECT: increased efficiency when encoding digital video signals and reduced speed of data transmission in bits.

4 cl, 8 dwg

FIELD: optics.

SUBSTANCE: method of optical registration of process, changing in time, is realized due to frame-by-frame shooting of digital camera with CCD array to get time dependence of at least one cinematic characteristic of tested process. During frame-by-frame shooting, registration is carried out in area of one or row of spatially fixed fragments of field of any frame. Video camera reads out information at any field from those columns or lines of CCD array only which are disposed in field of specified fragment. After that images of any registered fragment are combined to get time dependence in form of time-base.

EFFECT: improved reliability.

5 cl, 3 dwg

FIELD: method for interpolating values of sub-pixels during encoding and decoding of data.

SUBSTANCE: method of interpolation during video data encoding is claimed, which features an image, containing pixels ordered in rows and columns and represented by values having given dynamic range, where pixels in rows are in integral horizontal positions, and pixels in rows are in integral vertical positions, which image is interpolated in such a way that values of sub-pixels are generated in fractional horizontal and vertical positions, aforementioned method containing following stages: a) when values are required for sub-pixels in half-integral horizontal positions and integral vertical positions and in integral horizontal positions and half-integral vertical positions, such values are interpolated directly using weighted sums of pixels located in integral horizontal and integral vertical positions; b) when values are required for sub-pixels in half-integral horizontal positions and half-integral vertical positions, such values are interpolated directly using a weighted sum of values for sub-pixels located in half-integral horizontal positions and integral vertical positions, computed in accordance with stage a); and c), when values are required for sub-pixel in quaternary horizontal position and quaternary vertical position, such values are interpolated by averaging of at least one pair from first pair of values of sub-pixel located in half-integral horizontal position and half-integral vertical position, and of sub-pixel, located in integral horizontal position and integral vertical position, and second pair of values of pixel, located in integral horizontal position and integral vertical position, and of sub-pixel, located in semi-integral horizontal position and semi-integral vertical position.

EFFECT: creation of improved method for interpolating values of sub-pixels during encoding and decoding of data.

13 cl, 26 dwg, 2 tbl

FIELD: television.

SUBSTANCE: support frame is assigned with sign, showing information about direction of support frame, and during determining of predicted vector of movement of encoded block averaging operation is performed with use of vectors of movement of neighboring blocks, during which, if one of aforementioned blocks has movement vectors, information about direction of support frames is received, to which these movement vectors are related, and one of movement vectors is selected with reference to received information about direction, than averaging operation is performed with use of selected movement vector to receive subject movement vector of encoded block.

EFFECT: higher precision, higher reliability.

3 cl, 1 dwg, 3 ex

Up!