RU2371881C1 - Method of modelling video signal coding information for information compression/decompression - Google Patents

Abstract

FIELD: information technologies.

SUBSTANCE: there is proposed the method of modelling the context of video signal coding information for compression or decompression of coding information. Original function value for probabilistic coding of the coding information of the video signal of the improved level is determined on the basis of the coding information of the appropriate video signal of the main level.

EFFECT: providing the method of modelling the coding information context in order to increase data compression coefficient by using context adaptive binary arithmetic coding which is an entropic coding scheme of the improved video codec when the scalable coding scheme is combined with AVC MPEG-4.

6 cl, 7 dwg

Description

FIELD OF THE INVENTION

The present invention relates to a method for modeling video encoding information using entropy encoding for compression (decompression) or decompression (restoration) of encoding information.

State of the art

Scalable video coding (SVC) is a coding scheme that encodes video data into a sequence of images with the highest image quality, ensuring that part of the sequence of encoded images (in particular, a partial sequence of frames selected from a common sequence of frames) can be decoded and used to display video data with low image quality.

Despite the possibility of displaying video data with low image quality by receiving and processing part of a sequence of images encoded according to a scalable coding scheme, there remains the problem of significantly reducing image quality in the case of a low data rate. One solution to this problem is to provide an auxiliary sequence of images for low data rates, for example a sequence of images having a small screen size and / or low frame rate.

The auxiliary sequence of images is called the main level, and the main sequence of images is called the advanced level. When used in a decoding unit, the main sequence of images is also called the current level, decoded based on the main level. When providing a basic level to increase coding efficiency, inter-level prediction is performed.

SVC can be used in combination with the MPEG-4 video codec or the advanced MPEG-4 video codec (AVC), also called “H.264”. For adaptive application of binary arithmetic coding, which is an entropy coding scheme according to related coding information, there is a need to determine a method for encoding or decoding encoding data (e.g., syntax information) of a video codec.

Disclosure of invention

An object of the present invention is to provide a method for modeling a context of coding information to increase a data compression ratio using context adaptive binary arithmetic coding (CABAC), which is an AVC entropy coding scheme when a scalable coding scheme is combined with MPEG-4 AVC.

An object of the present invention can be achieved by providing a modeling method in which an initial value of a function for probabilistically encoding encoding information of a first level image block (e.g., a current level) is determined based on encoding information of a second level (e.g., a basic level) different from the first level.

In an embodiment of the present invention, flags that can be used to determine the initial function value for probabilistically encoding encoding information include a base_mode_flag flag indicating whether image data encoded in intra-frame encoding mode or block motion vector information is valid the second level corresponding to the image block should be used for the image block, the flag "base_mode_refinement_flag", indicating whether the block from it is necessary to improve the quality of the company to use the information of the motion vector of the second-level block corresponding to the image block, the flag "residual_prediction_flag" indicating whether the residual data of the image block was actually encoded using data predicted from the residual data of the second-level block corresponding to the image block, an intra_base_flag flag indicating whether the image data of the image block was actually encoded into the difference data based on the image data, code mvd_ref_1X flag indicating the quality improvement value required to obtain the motion vector of the image block using the information of the motion vector of the second level block corresponding to the image block, as well as the flag "motion_prediction_flag_1X" indicating whether the motion vector of the second level block corresponding to the image block is really to be used for the predictive motion vector of the image block agony.

In another embodiment of the present invention, encoding information of a second level block corresponding to an image block is used as second level encoding information. And also the encoding information of the second level block corresponds to the encoding information of the image block.

In another embodiment of the present invention, information indicating whether a second level block corresponding to the image block is encoded in inter-frame coding mode or in int-frame coding mode is used as second-level coding information.

In another embodiment of the present invention, the initial value is determined based on whether the segmentation information of the second level block corresponding to the image block is identical to the segmentation information of the image block.

In another embodiment of the present invention, the initial value is determined either based on the value of the quantization parameter of the second level block corresponding to the image block, or based on the difference between the value of the quantization parameter of the second level block and the value of the quantization parameter of the image block.

In another embodiment of the present invention, the initial value is determined based on the difference between the motion vector of the image block and the motion vector of the second level block corresponding to the image block.

In another embodiment of the present invention, the initial value is determined based on whether the value indicating the reference image of the image block is identical to the value indicating the reference image of the second level block corresponding to the image block.

In another embodiment of the present invention, the initial value is determined based on whether the spatial resolution of the image block is identical to the spatial resolution of the second level block corresponding to the image block.

In another embodiment of the present invention, block structure information indicating whether a value other than 0 is actually present in the second level block corresponding to the image block is used as second level encoding information.

In another embodiment of the present invention, to determine the initial value, based on the second level encoding information, two constants “m” and “n” are selected, based on the selected constants “m” and “n”, an intermediate value and information related to the encoding of the second level (for example, the value of the quantization parameter of the second level, the difference between the values of the quantization parameter of the first and second levels or the ratio of spatial resolution between the first and second levels), and based on ity whether intermediate value is greater than a predetermined value, determined initial probability value MPS and coding information.

Brief Description of the Drawings

Figure 1 depicts a block diagram of a block execution unit CABAC advanced level coding unit performing context modeling, according to the present invention;

Figure 2 depicts an example of binarization of the encoding input information;

Figure 3 depicts a context modeling method for probabilistically encoding encoding information;

Figure 4 depicts a routine for determining the initial value for probabilistic coding; and

5 is a block diagram of a CABAC execution unit of an advanced layer decoding unit performing context modeling in accordance with the present invention;

6 depicts block structure information according to the present invention; and

7 depicts a method for performing context modeling using block structure information and residual prediction indicator information according to the present invention.

Preferred Embodiment

Next, detailed reference will be made to preferred embodiments of the present invention, examples of which are illustrated in the drawings.

FIG. 1 is a block diagram of an advanced level CABAC execution unit according to a preferred embodiment of the present invention. The CABAC execution unit shown in FIG. 1 includes a binarization unit 101, a context modeling unit 102, and an arithmetic coding unit 110. The binarization unit 101 performs binarization of the incoming non-binary coding information according to a specific pattern. In particular, the binarization unit 101 converts a non-binary syntax element into a binary string, as shown in FIG. The context modeling unit 102 simulates each bit of binary encoding information not only on the basis of encoding information of a neighboring block of the same level (in this example, an advanced level), but also on the basis of encoding information of the main level correlated with binary encoding information or inter-level information 10 correlation. Arithmetic coding unit 110 performs arithmetic coding of an input bit based on an established model.

Arithmetic coding unit 110 includes a standard coding mechanism 103 for performing arithmetic coding of bits of coding information based on variables (in particular, probabilistic functions and initial values of probabilistic functions) modeled by context modeling unit 102, as well as bypass coding mechanism 104 for performing arithmetic coding encoding information that does not benefit from the simulation, since bits 0 and 1 of the encoding information are almost similar th the probability of occurrence.

The present invention, which uses a simulation of coding input information, is not directly related to an arithmetic coding procedure based on simulated variables. Therefore, the description of bit compression (entropy encoding) associated with the arithmetic encoding unit 110 is omitted herein, since this is not necessary for understanding the present invention.

If the incoming encoding information has a non-binary value, then the CABAC execution unit shown in FIG. 1 binarizes the value of the incoming encoding information using the binarization unit 101. Figure 2 depicts an example of binarization. The coding information used in the example shown in FIG. 2 is associated with macroblock types (mb_type). The types of macroblocks (with direct coding, with intraframe coding, P_16 × l6, P_16 × 8, P_8 × l6 and P_8 × 8) are assigned the corresponding binary values according to a predefined scheme (or conversion table). Other coding information is binarized according to other schemes (or another conversion table) defined for the respective elements in a manner similar to that shown in FIG. 2.

In bit compression, bits obtained using the aforementioned binarization are input to the arithmetic coding unit 110. Bits of encoding information having a similar probability of bit values 0 and 1 occurring in the encoding information are directly input to the bypass encoding mechanism 104, while bits of encoding information having other probabilities of bit values 0 and 1 occurring in the encoding information are input to block 102 modeling the context so that the incoming bits undergo a simulation process.

The context modeling unit 102 performs bit modeling of the incoming encoding information at an advanced level, either based on the encoding information correlated with the bit values of the corresponding encoding information of the neighboring macroblock and / or the value of the corresponding encoding information received from the main layer encoding unit (not shown), or based on information 10 related to the relationship between the advanced level and the main level. Modeling is the process of selecting a probability function and determining the initial value of the probability function. As shown in FIG. 3, the offset value (..., k-1, k, k + 1, ...) is determined, according to the encoding information, for selecting the probability function (..., f _k-1 , f _k , f _{k + 1} , ...) coding information, and the value of the index variable "ctxIdxInc" is determined from the offset value according to information correlated with the coding information. “CtxIdxInc” is defined as the value of the index variable, and the initial values “valMPS” and “pStateIdx” are determined for use with the probability function. As the initial value, “pStateIdx” is determined, and the initial probability of LPS (or MPS) is determined by the method shown in FIG. 3. Accordingly, the standard encoding mechanism 103 performs encoding (or compression) of the bits of the incoming encoding information using the selected probabilistic function, starting with the determined initial values “valMPS” and “pStateIdx”.

The following are detailed examples of a method for determining the value of the index variable “ctxIdxInc” in the simulation performed by the context modeling unit 102. Many of the examples below are simply examples of a method for modeling specific coding information of an advanced layer, either based on coding information correlated with the value of certain coding information or based on the relationship between the advanced and the core layer. In this regard, the present invention is not limited to the examples below, and any method characterized by modeling the coding information of the advanced level, either based on the coding information correlated with the value of the coding information element, or based on the relationship between the advanced and the basic levels, is within the scope of the present invention.

First, a plurality of methods are described for determining the index variable “ctxIdxInc” of the flag “base_mode_flag”, indicating whether the encoding information (such as motion vector information or image data encoded in the intra-frame encoding mode) of the base layer block corresponding to any macroblock should be used for the macroblock.

1-1) ctxIdxInc = condTermFlagA + condTermFlagB + condTermFlagBase

In this case, “A” and “B” denote neighboring macroblocks located on the upper and left sides of the current macroblock X. The flag “condTermFlagN” (N = A or B) is set to “0” if macroblock N is not available, or in if the flag "base_mode_flag" of the macroblock N is assigned the value "0", otherwise it is assigned the value "1". Similarly, the “condTermFlagBase” flag is assigned the value “0” if there is no main-level block corresponding to the current macroblock X, or if the “base_mode_flag” flag of the corresponding block is set to “0”, otherwise it is assigned the value “1”. In this regard, the value of the corresponding coding information of the main level is used as the basis for determining the value of the index variable "ctxIdxInc". It indicates that the initial value for probabilistic coding changes depending on the value of the corresponding coding information of the main level.

1-2) ctxIdxInc = condTermFlagA '+ condTermFlagB' + condTermFlagBase '

In this case, the “conTermFlag” flag of the block is assigned the value “0” or “1” depending on which of the modes (inter-frame coding or int-frame coding) the block is encoded. For example, the flag “condTermFlagBase '” is assigned the value “0” (or “1”) if the main-level block corresponding to the current macroblock X is encoded in inter-frame coding mode, and the value “1” (or “0”) is assigned to if the corresponding block is encoded in intra-frame coding mode.

In this method, the modes (interframe or intraframe coding) of two neighboring blocks A and B, as well as the corresponding block of the main level, are used as the basis for determining the initial value of the probability function for encoding the flag bit "base_mode_flag".

Alternatively, only the “condTermFlagBase 'flag” (that is, only the mode of the corresponding main level block) can be used as the basis for determining the value of the “ctxIdxInc” index variable to change the initial value depending solely on the value of the “condTermFlagBase” flag.

1-3) ctxIdxInc = (BaseBlkSize == EnhanceBlkSize)? 1: 0 + condTermFlagA + condTermFlagB

In this method, a value indicating whether the segment of the main layer block is identical to the segment of the advanced level block (for example, a value of “1” in the case of identity and a value of “0” in the absence of identity), or a value indicating whether or not the block size of the advanced level is equal to the size of the corresponding block of the main level (for example, the value “1” in case of equality and the value “0” in case of inequality) is used as the basis for determining the initial value of Local function.

1-4) ctxIdxInc = condTermFlagA "+ condTermFlagB" + condTermFlagBase "

In this case, the “condTermFlag” flag of the block is assigned the value “1” if the quantization parameter of the block has a value equal to or greater than a predefined threshold, otherwise it is assigned the value “0.” In this method, the quantization parameters of two neighboring blocks A and B, as well as the quantization parameter of the corresponding block of the main level, are used as the basis for determining the initial value of the probability function for encoding the flag bit “base_mode_flag”.

The “condTermFlag” flag of the block can also be set to “1” or “0” depending on the difference between the value of the quantization parameter of the block and the value of another quantization parameter, instead of depending on the quantization parameter of the block, to determine the value of the index variable “ctxIdxInc” based on flag values "condTermFlag" ". For example, the flag “condTermFlagN” of block N is assigned the value “1” if the difference between the value of the quantization parameter of block N and the value of the quantization parameter of the block of the main level corresponding to block N is equal to or greater than a predetermined threshold, otherwise the flag “condTermFlagN "" Is assigned the value "0". In this example, “condTermFlagBase” ”indicates a flag indicating whether the difference between the quantization parameter value of the current block X and the quantization parameter value of the base layer block corresponding to the current block X is greater or less than a predetermined threshold.

Alternatively, only the flag “condTermFlagBase” ”(that is, only the value of the quantization parameter of the block of the main level corresponding to the current block X (or only the difference between the value of the quantization parameter of the current block X and the value of the quantization parameter of the corresponding block)) can be used as the basis for determining the value of the index variable "ctxIdxInc" to change the initial value depending only on the value of the flag "condTermFlagBase" ".

1-5) ctxIdxInc = 0 (if C> threshold 1),

1 (if threshold 1> C> threshold 2),

2 (if C <threshold 2)

In this case, “C” denotes either the motion vector of the corresponding block of the main level, or the difference between the motion vector of the corresponding block and the motion vector of one of the neighboring macroblocks, or the average value of the motion vector of the neighboring macroblocks.

In this regard, the motion vector of the main level is used as the basis for determining the initial value of the probability function.

1-6) CtxIdxInc = (refIdx _Enhance L1 == refIdx _Base L1)? 1: 0+ (refIdx _Enhance L0 == refIdx _Base L0)? 1-0

In this method, a value indicating whether the indices of the reference image refIdxL0 and refIdxL1 of the image groups L0 and L1 of the macroblock containing the currently encoded information are equal to the indexes of the reference images of the image groups L0 and L1 of the corresponding block of the main level (for example, the value "2 "If the indices of the reference image refIdxL0 and refIdxL1 of the image groups L0 and L1 are equal to the indices of the reference image of the main level, the value is" 1 "if one of the indices of the reference image refIdxL0 and refIdxL1 is the corresponding index of the reference image of the main level, and the value “0” if none of the indices of the reference image refIdxL0 and refIdxL1 is equal to the index of the reference image of the main level), is used as the basis for determining the initial value of the probability function for changing the initial value depending from a value indicating whether the indices of the reference image of the advanced level are equal to the indices of the reference image of the main level.

A combination of the above methods (from 1-1 to 1-6) instead of one of the above methods can be used to determine the initial value of the probability function for the entropy coding of the flag "base_mode_flag".

The following describes many methods for determining the index variable "ctxIdxInc" of the flag "base_mode_refinement_flag", indicating whether any macroblock really needs to improve the quality to use the motion vector information of the main level block corresponding to the macroblock.

Since the base_mode_refinement_flag flag is not used when encoding the corresponding main layer macroblock in intra-frame encoding mode, a method involving intra-frame encoding mode, such as a method similar to the above method 1-2), is not used to simulate the base_mode_refinement_flag flag bit.

2-1) ctxIdxInc = condTermFlagA + condTermFlagB + condTermFlagBase

The “condTermFlagN” flag (N = A or B) is set to “0” if the macroblock N is not available, or if the “base_mode_refinement_flag” flag of the macroblock N is set to “0”, otherwise it is set to “1” . Similarly, the flag “condTermFlagBase” is assigned the value “0” if there is no main-level block corresponding to the current macroblock, or if the flag “base_mode_refinement_flag” of the corresponding block is assigned the value “0”, otherwise it is assigned the value “1”. In this regard, the value of the corresponding coding information of the main level is used as the basis for determining the value of the index variable "ctxIdxInc".

2-2) ctxIdxInc = (BaseBlkSize == EnhanceBlkSize)? 1: 0 + condTermFlagA + condTermFlagB

This method is similar to the above method 1-3).

2-3) ctxIdxInc = condTermFlagA "+ condTermFlagB" + condTermFlagBase "

or ctxIdxInc = condTermFlagBase "

This method is similar to the above method 1-4).

2-4) ctxIdxInc = (SpatialRes _Enhance == SpatialRes _Base )? 1-0

In this method, a value indicating whether the spatial resolution of the basic level image is equal to the spatial resolution of the advanced level image (for example, the value “1” in case of equality and the value “0” in case of inequality) is used as the basis for determining the initial value probabilistic function.

A combination of the above methods (from 2-1 to 2-4) instead of one of the above methods can be used to determine the initial value of the probability function for probabilistic coding of the flag "base_mode_refinement_flag".

Next, a description will be given of a plurality of methods for determining the index variable “ctxIdxInc” of the “residual_prediction_flag” flag, indicating whether the residual data of any macroblock was actually encoded using data predicted from the residual data of the base layer block corresponding to the macroblock.

3-1) ctxIdxInc = condTermFlagA + condTermFlagB + condTermFlagBase

The “condTermFlagN” flag (N = A or B) is set to “0” if the macroblock N is not available, or if the “residual_prediction_flag” flag of the macroblock N is set to “0”, otherwise it is set to “1” . Similarly, the flag “condTermFlagBase” is assigned the value “0” in the absence of a block of the main level corresponding to the current macroblock, or if the flag “residual_prediction_flag” of the corresponding block is assigned the value “0”, otherwise it is assigned the value “1”. In this regard, the value of the corresponding coding information of the main level is used as the basis for determining the value of the index variable "ctxIdxInc".

3-2) ctxIdxInc = (BaseBlkSize == EnhanceBlkSize)? 1: 0 + condTermFlagA + condTermFlagB

This method is similar to the above method 1-3).

3-3) ctxIdxInc = condTermFlagA "+ condTermFlagB" + condTermFlagBase "

or ctxIdxInc = condTermFlagBase "

This method is similar to the above method 1-4).

3-4) ctxIdxInc = (refIdx _Enhance L1 == refIdx _Base L1)? 1: 0+ (refIdx _Enhance L0 == refIdx _Base L0)? 1-0

This method is similar to the above method 1-6).

3-5) ctxIdxInc = 0 (if C> threshold 1),

1 (if threshold 1> C> threshold 2),

2 (if C <threshold 2)

This method is similar to the above method 1-5).

3-6) ctxIdxInc = (SpatialRes _Enhance == SpatialRes _Base )? 1-0

This method is similar to the above method 2-4).

3-7) ctxIdxInc = CBP _Base ? 1-0

In this method, the initial value of the probability function for encoding the flag "residual_prediction_flag" is determined from the composition value of the encoded block (CBP) of the corresponding block of the main level. In this case, the CBP block of luminance (luminance) or block of chrominance (chrominance) of the main level can be used as a CBP. The CBP of the block is assigned a value other than "0", if a bit value other than "0" is present in the block, otherwise it is assigned the value "0". In this method, the initial value of the probability function for encoding the flag "residual_prediction_flag" is set differently depending on the value indicating whether the corresponding block of the main level actually contains a value other than "0", which is set to "1" if the value is present other than “0” and is set to “0” if there is no value other than “0”.

In the method of using CBP to determine the initial value of the corresponding coding information "residual_prediction_flag" of neighboring blocks A and B, in addition to the above condition (CBP _Base ? 1: 0), can be used as conditions for determining the index variable "ctxIdxInc". In this case, the index variable "ctxIdxInc" can be defined as follows:

ctxIdxInc = CBP _Base ? 1: 0 + condTermFlagA + condTermFlagB

Alternatively, the index variable "ctxIdxInc" can be determined based on the CBP values of two neighboring blocks A and B as follows:

ctxIdxInc = CBP _A ? 1: 0 + CBP _B ? 1-0

A combination of the above methods (from 3-1 to 3-7) instead of one of the above methods can be used to determine the initial value of the probability function for probabilistic coding of the flag "residual_prediction_flag".

Modeling (for example, setting the initial value) of the encoding information, in addition to the encoding information described above, can also be performed in various ways, either according to the encoding information of the main level, or according to the inter-level relationship.

For example, a simulation for the probabilistic coding of the intra_base_flag flag indicating whether the image data of the advanced layer block has actually been encoded into difference data based on image data encoded in the intraframe coding mode of the base layer block corresponding to the advanced level block can also be performed in various ways or using an inter-level relation (in particular, using the corresponding coding information of the main level) as As with a method similar to method 1-1), either using an inter-level spatial resolution ratio according to a method similar to method 2-4), or using a quantization parameter that reflects the image quality level of the main level, according to a method similar to method 1-4).

In addition, modeling for the probabilistic coding of the information “mvd_ref_1X, X = 0.1”, reflecting the value of the quality improvement required by any macroblock to use the motion vector information of the main level block corresponding to the macroblock, can also be performed in various ways or using the inter-level relation (in in particular, using the corresponding coding information of the main level) according to a method similar to method 1-1), or using the inter-level spatial resolution ratio but a method similar to 2-4).

In addition, the simulation for the probabilistic coding of the flag “motion_prediction_flag_1X, X = 0.1”, indicating whether the motion vector of the main level block corresponding to the macroblock really should be used for the predictive motion vector of the macroblock can also be performed in various ways or using inter-level relation (in particular, using the corresponding coding information of the main level) according to a method similar to method 1-1), or using the inter-level relation of the spatial time Addressing according to a method similar to 2-4), or the ratio of block sizes according to a method similar to 1-3).

Modeling of three types of information (intra_base_flag, mvd_ref_1X, motion_prediction_flag_1X) can also be performed using other inter-level relations that differ from the above relations.

Many of the above modeling methods can also be applied to any other encoding information, the value of which can affect the inter-level relationship.

Although the above description was made with reference to the fact that the initial values “valMPS” and “pStateIdx” are determined from the index variable “ctxIdxInc”, these two initial values are determined from the values “m” and “n”, which are determined from the index variable "ctxIdxInc" depicted in figure 4 the way.

The intermediate value "preCtxState" in the routine for determining the initial value shown in Figure 4 is determined using the function "Clip3 ()". The “Clip3 ()” function of the “preCtxState” definition contains the “SliceQPY” parameter for quantizing brightness as an argument (varX) in addition to the “m” and “n” values. This argument (varX) is associated with a macroblock containing the currently encoded encoding information. The argument (varX), which affects the definition of the initial value, in addition to the values of "m" and "n", does not have a value associated with the inter-level relationship.

Accordingly, if the inter-level relationship is reflected in the argument (varX) to obtain the initial value based on the inter-level relationship, then the initial value is likely to have a value that is much more beneficial with respect to probabilistic coding. In this regard, according to the present invention, the inter-level relationship is reflected in the argument (varX).

The methods for reflecting the inter-level relationship in the argument (varX) should either use the BaseSliceQPY parameter for quantizing the brightness of the main level as an argument (varX), or use the difference between the quantization parameters of the advanced and main levels as an argument (varX), or use the spatial resolution coefficient between main level and current level as an argument (varX).

If the difference between the quantization parameters of the advanced and the basic levels is used as an argument (varX), then the preCtxState function 401 depicted in FIG. 4 according to the present invention can be determined as follows:

preCtxState = Clip3 (1,126, ((m * (SliceQPYBaseSliceQPY)) >> 4) + n)

Although the aforementioned method has been described for probabilistic coding in a coding unit, a similar method is applied to the CABAC decoding unit shown in FIG. 5 for decompressing compressed data. In this regard, the description of the context modeling in the decoding unit is omitted herein.

Similar to the method in which the context modeling unit 102 of the CABAC encoding unit shown in FIG. 1 models the encoding target information, the context modeling unit 102 of the CABAC decoding unit shown in FIG. 5 models the encoding target information based on the core layer encoding information and the inter-layer ratio information 20, and also transmits the corresponding initial value to the standard decoding mechanism 203 provided next to the context modeling unit 202. The standard decoding engine 203 converts the bits of the incoming encoding information into a compressed bit string starting with a value similar to the original value used in the standard encoding mechanism 103 of the encoding block.

6 depicts elements of encoded block composition information (CBP) according to the present invention.

The block structure information is a type of characteristic bit flag indicating whether a residual is indeed present, in particular whether an 8 × 8 luminance residual block is actually present. For example, if the main macroblock (baseMB) corresponding to the macroblock (currMB) is present in the current frame, then the main macroblock is divided into four blocks, and the zero bit of the block structure information indicates whether there is a remainder in the upper left block, the first bit indicates then, is there really a remainder in the upper right block, the second bit indicates whether there is really a remainder in the lower left block, and a third bit indicates whether a remainder is really in the lower right block. Block structure information is displayed using six bits, including 0-3 bits and two bits (AC and DC), representing color blocks. In this regard, the structure information of the block indicates whether a residue is actually present in each of the divided blocks, and the presence of a residue is indicated by a value of “1”, and the absence of a residue is indicated by a value of “0”. In addition, the initial value of the probability function for encoding the residual prediction flag information (residual_prediction_flag) is set differently depending on whether a value other than “0” is actually present in the corresponding lower level block.

7 depicts a method for performing context modeling using block structure information and residual prediction flag information according to the present invention.

In an embodiment of the present invention, it is assumed that the residual prediction flag information is assigned a value of “1” at both levels 1 and 2 (Layer1 and Layer2), and the block composition information is assigned a value of “0” at level 1. In this case, the level 1 encoding information is not can be used when modeling the context of level 2. This is due to the fact that the information of the composition of the block of level 1 is assigned the value "0", indicating that there is no remainder in the block of level 1 corresponding to the macroblock of level 2 OK. However, it can be seen that a remainder is present because the value of the structure information of the level 0 block is greater than “0”. It follows that, even if the information of the structure of the level 1 block is assigned the value “0”, the information of the prediction flag of the residuals of level 1 should also be taken into account to determine the actual presence of the remainder. Accordingly, the present invention uses an upsampled residual of level 0, which is below level 1, when predicting residuals of level 2.

Next, a description will be given of another embodiment of the present invention, which provides many methods for determining the index variable “ctxIdxInc” of the residual_prediction_flag residual prediction flag information indicating whether residual data of any macroblock has actually been encoded using data predicted from the residual data of the main block level corresponding to the macroblock.

ctxIdxInc = min (1, resPredFlag _base + CBP _base )

Equation 1

The method expressed by equation (1) is provided for selecting either the value “0” or the value “1” as the state of the index variable. If the residual prediction flag information of the corresponding macroblock of the main frame is set to “0”, then the residual prediction flag information (resPredFlag _base ) of the main frame is assigned the value “0”, otherwise it is assigned the value “1”. Similarly, the value “0” or “1” is assigned to the CBP _{base information of the} main frame. In this regard, the sum of the residual prediction flag information value (resPredFlag _base ) and the block structure information value (CBP _base ) has a value of “0”, “1” or “2”. The index variable has a value of “0” or “1” because the less of “1” and the sum value (“0”, “1” or “2”) is assigned to the index variable.

ctxIdxInc = resPredFlag _base + CBP _base

Equation 2

The method expressed by equation (2) is provided for selecting either the value “0”, or the value “1”, or the value “2” as the state of the index variable.

The following is a description of another embodiment of the present invention that provides a method for determining an index variable (ctxIdxInc) using the sum of the residual information values of the corresponding blocks of the main frame. This embodiment can be applied if the blocks of the current frame do not match the corresponding blocks of the main frame, that is, in the presence of a plurality of corresponding main macroblocks.

Equation 3

In equation (3), “P _baseij ” denotes the pixel value at position (i, j) of the residual information of the corresponding block of the main image. “Energy _base ” indicates the presence of a residue, which can be determined by searching all the pixels of the corresponding block of the main image.

ctxIdxInc = Energy _base ? 1-0

Equation 4

In the method expressed by equation (4), all pixels are searched for the presence of the image value, and the value “1” is assigned to the index variable in the presence of any pixel value, and the value “0” is assigned to the index variable in the absence of any pixel value.

0 Energy _base <= threshold 1

ctxIdxInc = 1 threshold 1 <Energy _base <threshold 2

2 Energy _base > = threshold 2

Equation 5

Expressed by equation (5), the method uses many thresholds (threshold 1 and threshold 2) so that the value “0” is assigned to the index variable if the total sum of the pixel values is less than or equal to the threshold (threshold 1), the value “1” is assigned to the index variable if the total amount is between the thresholds (threshold 1 and threshold 2), and the value “2” was assigned to the index variable if the total amount is equal to or greater than the threshold (threshold 2).

A decoding unit including a context modeling unit for modeling encoding information according to the above method may be included in a mobile communication terminal, media player, or the like.

Although the invention has been described with reference to preferred embodiments, those skilled in the art will understand that various improvements, modifications, replacements and additions can be made to the invention without departing from the spirit and scope of the invention. Thus, this means that the invention encompasses improvements, modifications, replacements and additions within the scope of the appended claims and their equivalents.

Industrial applicability

As described above, with reference to limited embodiments, the present invention simulates the context of each bit of encoding information using an inter-layer relation to determine the initial value of a function for probabilistically encoding encoding information, which is preferred with respect to probabilistic encoding, (i.e., the initial value , further reducing the initial probability of LPS), thus significantly increasing the coefficient of com Ressam probabilistic data encoding.

Claims (6)

1. A method for decoding a video signal, comprising: receiving a video signal including an advanced layer bit stream and a main layer bit stream; analyzing the first encoding information based on the second encoding information indicating whether the prediction information of the current block at the advanced level is extracted from the corresponding block at the main level, the first encoding information indicates whether the residual data of the current block at the advanced level is predicted from the corresponding block at the main level ; and decoding the video signal based on the first encoding information, wherein the analysis step further includes: obtaining a context index offset of the first encoding information; extracting context index increment information based on the second encoding information; obtaining context index information by adding the context index offset and the context index increment information; determining the initial value of the probability function based on the context index information; and performing arithmetic coding on the first coding information based on the initial value of the probability function. 2. The method of claim 1, wherein the prediction information includes at least one of macroblock type information and motion information. 3. The method according to claim 1, in which the first level differs from the second level in quality and / or spatial resolution. 4. The method according to claim 3, in which the second level has a lower quality than the first level. 5. The method according to claim 1, wherein the initial value includes state probability information and the most probable value information. 6. A video decoding apparatus comprising a video signal receiving unit receiving a video signal including an advanced level bitstream and a main level bitstream; a context adaptive binary arithmetic coding (SAVAS) execution unit analyzing the first encoding information based on the second encoding information indicating whether the prediction information of the current block at an advanced level is extracted from the corresponding block at the main level, the first encoding information indicates whether residual data of the current block at the advanced level from the corresponding block at the main level; and an advanced decoder decoding the video signal based on the first encoding information, wherein the SAVAC execution unit includes a context modeling unit that: obtains a context index offset of the first encoding information and extracts context index increment information based on the second encoding information; obtains context index information by adding the context index offset and the context index increment information; determines the initial value of the probability function based on the context index information; and performs arithmetic coding on the bits of the first coding information based on the initial value of the probability function.

Images