KR102071665B1 - Method and apparatus for parallel entropy encoding/decoding - Google Patents

Abstract

The entropy decoding method according to the present invention includes deriving scheduling information from a header of a bitstream received from an encoder, and using the derived scheduling information, inverse binarization of a partial bitstream constituting the received bitstream. reordering according to de-binarization order, deriving a bin corresponding to the bitstream based on the reordered partial bitstream, and debinarizing the derived bin to obtain a syntax element. Steps. According to the present invention, image encoding / decoding efficiency can be improved.

Description

Parallel Entropy Encoding / Decoding Method and Apparatus {METHOD AND APPARATUS FOR PARALLEL ENTROPY ENCODING / DECODING}

The present invention relates to image processing, and more particularly, to an entropy encoding / decoding method and apparatus.

Recently, as the broadcasting service having high definition (HD) resolution has been expanded not only in Korea but also in the world, many users are getting used to high resolution and high quality video, and many organizations are accelerating the development of next generation video equipment. In addition, as the interest in Ultra High Definition (UHD), which has more than four times the resolution of HDTV, is increasing, a compression technology for higher resolution and high quality images is required.

For image compression, an inter prediction technique for predicting a pixel value included in a current picture from a previous and / or subsequent picture in time, and for predicting a pixel value included in a current picture using pixel information in the current picture. An intra prediction technique, an entropy encoding technique of allocating a short code to a symbol with a high frequency of appearance and a long code to a symbol with a low frequency of appearance may be used.

An object of the present invention is to provide an image encoding method and apparatus capable of improving image encoding / decoding efficiency.

Another object of the present invention is to provide an image decoding method and apparatus for improving image encoding / decoding efficiency.

Another object of the present invention is to provide an entropy encoding method and apparatus capable of improving image encoding / decoding efficiency.

Another technical problem of the present invention is to provide an entropy decoding method and apparatus for improving image encoding / decoding efficiency.

1. An embodiment of the present invention is an entropy decoding method. The method includes deriving scheduling information from a header of a bitstream received from an encoder, and using the derived scheduling information, de-binarization of a partial bitstream constituting the received bitstream. Reordering according to the order, deriving a bin corresponding to the bitstream based on the reordered partial bitstream, and inversely binarizing the derived bin to obtain a syntax element Wherein the scheduling information includes information about a location of each of the partial bitstreams constituting the received bitstream and information about an inverse binarization order of the partial bitstream.

2. The method of 1, wherein the total probability intervals for the probability of the Least Probable Bin (LPB) of the bin are classified into a plurality of probability intervals each having a representative probability, and the partial bitstream includes at least one codeword. Codewords constituting the partial bitstream may correspond to the same representative probability.

3. The method of claim 2, wherein the deriving of the bin comprises: deriving codeword length information from a header of the received bitstream and deriving a bin corresponding to the bitstream based on the derived codeword length information. The codeword length information may further include information regarding a minimum length and a maximum length of a codeword for a representative probability of each of the plurality of probability intervals.

4. The method of 3, wherein the bin derivation step comprises: determining a codeword-bin mapping table for a representative probability of each of the plurality of probability intervals using the derived codeword length information and the determined codeword. The method may further include deriving a bin corresponding to the bitstream using the bin mapping table.

5. Another embodiment of the present invention is an entropy decoding apparatus. The apparatus derives scheduling information from a header of a bitstream received from an encoder, and de-binarization order of the partial bitstream constituting the received bitstream using the derived scheduling information. A bitstream determiner reordering according to the present invention, a bin decoder that derives bins corresponding to the bitstreams based on the rearranged partial bitstreams, and inverse binarizes the derived bins. a de-binarizer for acquiring a syntax element, wherein the scheduling information includes information about a position of each of the partial bitstreams constituting the received bitstream and an inverse binarization order of the partial bitstream. Include information about

6. The method of claim 5, wherein the total probability intervals for the probability of the Least Probable Bin (LPB) of the bin are classified into a plurality of probability intervals each having a representative probability, and the partial bitstream includes at least one codeword. Codewords constituting the partial bitstream may correspond to the same representative probability.

7. The method of 6, further comprising: a codeword length calculator for deriving codeword length information from a header of the received bitstream, wherein the empty decoder corresponds to the bitstream based on the derived codeword length information. The bin may be derived, and the codeword length information may be information about a minimum length and a maximum length of a codeword for a representative probability of each of the plurality of probability intervals.

8. The method of claim 7, further comprising a mapping table determiner for determining a codeword-bin mapping table for the representative probability of each of the plurality of probability intervals using the derived codeword length information. The decoder may derive a bin corresponding to the bitstream using the determined codeword-bin mapping table.

9. Another embodiment of the present invention is a video decoding method. The method includes deriving scheduling information from a header of a bitstream received from an encoder, and using the derived scheduling information, de-binarization of a partial bitstream constituting the received bitstream. Reordering according to the order, deriving a bin corresponding to the bitstream based on the reordered partial bitstream, and inversely binarizing the derived bin to obtain a syntax element And generating a reconstructed image by using the acquired syntax element, wherein the scheduling information includes information about a position of each partial bitstream constituting the received bitstream and an inverse binarization order of the partial bitstream. Contains information about

10. The method of claim 9, wherein the total probability intervals for the probability of the Least Probable Bin (LPB) of the bin are classified into a plurality of probability intervals each having a representative probability, and the partial bitstream includes at least one codeword. Codewords constituting the partial bitstream may correspond to the same representative probability.

11. The method of claim 10, wherein the bin derivation step comprises: deriving codeword length information from a header of the received bitstream and deriving a bin corresponding to the bitstream based on the derived codeword length information. The codeword length information may further include information regarding a minimum length and a maximum length of a codeword for a representative probability of each of the plurality of probability intervals.

12. The method of claim 11, wherein the bin derivation step comprises: determining a codeword-bin mapping table for a representative probability of each of the plurality of probability intervals using the derived codeword length information and the determined codeword. The method may further include deriving a bin corresponding to the bitstream using the bin mapping table.

According to the image encoding method according to the present invention, the image encoding / decoding efficiency can be improved.

According to the image decoding method according to the present invention, the image encoding / decoding efficiency can be improved.

According to the entropy encoding method according to the present invention, image encoding / decoding efficiency can be improved.

According to the entropy decoding method according to the present invention, image encoding / decoding efficiency can be improved.

1 is a block diagram illustrating a configuration of an image encoding apparatus according to an embodiment of the present invention.
2 is a block diagram illustrating a configuration of an image decoding apparatus according to an embodiment of the present invention.
3 is a table schematically illustrating one embodiment of a codeword length restriction according to the present invention.
4 is a conceptual diagram schematically showing an embodiment of a scheduling method for a bitstream parsing process according to the present invention.
5 is a conceptual diagram schematically showing another embodiment of a scheduling method for a bitstream parsing process according to the present invention.
6 is a block diagram schematically illustrating an embodiment of a parallel entropy encoding apparatus according to the present invention.
7 is a block diagram schematically illustrating an embodiment of a parallel entropy decoding apparatus according to the present invention.
8 is a block diagram schematically showing another embodiment of a parallel entropy encoding apparatus according to the present invention.
9 is a block diagram schematically showing another embodiment of a parallel entropy decoding apparatus according to the present invention.
10 is a block diagram schematically showing another embodiment of a parallel entropy encoding apparatus according to the present invention.
11 is a block diagram schematically showing another embodiment of a parallel entropy decoding apparatus according to the present invention.
12 is a flowchart schematically illustrating an embodiment of a parallel entropy encoding method according to the present invention.
13 is a flowchart schematically showing an embodiment of a parallel entropy decoding method according to the present invention.

EMBODIMENT OF THE INVENTION Hereinafter, embodiment of this invention is described concretely with reference to drawings. In describing the embodiments of the present specification, when it is determined that a detailed description of a related well-known configuration or function may obscure the gist of the present specification, the detailed description thereof will be omitted.

When a component is said to be “connected” or “connected” to another component, it may be directly connected to or connected to that other component, but it may be understood that another component may exist in between. Should be. In addition, the description "include" a specific configuration in the present invention does not exclude a configuration other than the configuration, it means that additional configuration may be included in the scope of the technical spirit of the present invention or the present invention.

Terms such as first and second may be used to describe various components, but the components should not be limited by the terms. The terms are used only for the purpose of distinguishing one component from another. For example, without departing from the scope of the present invention, the first component may be referred to as the second component, and similarly, the second component may also be referred to as the first component.

In addition, the components shown in the embodiments of the present invention are independently shown to represent different characteristic functions, and do not mean that each component is made of separate hardware or one software unit. In other words, each component is included in each component unit for convenience of description, and at least two components of each component may be combined into one component, or one component may be divided into a plurality of components to perform a function. Integrated and separate embodiments of the components are also included within the scope of the present invention without departing from the spirit of the invention.

In addition, some of the components may not be essential components to perform essential functions in the present invention, but may be optional components for improving performance. The present invention can be implemented including only the components necessary to implement the essentials of the present invention, except for the components used for improving performance, and a structure including only essential components except for the optional components used for improving performance. Also included in the scope of the present invention.

1 is a block diagram illustrating a configuration of an image encoding apparatus according to an embodiment of the present invention.

Referring to FIG. 1, the image encoding apparatus 100 may include a motion predictor 111, a motion compensator 112, an intra predictor 120, a switch 115, a subtractor 125, and a converter 130. And a quantization unit 140, an entropy encoding unit 150, an inverse quantization unit 160, an inverse transform unit 170, an adder 175, a filter unit 180, and a reference picture buffer 190.

The image encoding apparatus 100 may encode an input image in an intra mode or an inter mode and output a bitstream. In the intra mode, the switch 115 may be switched to intra, and in the inter mode, the switch 115 may be switched to inter. The image encoding apparatus 100 may generate a prediction block for an input block of an input image and then encode a residual between the input block and the prediction block.

In the intra mode, the intra predictor 120 may generate a prediction block by performing spatial prediction using pixel values of blocks that are already encoded around the current block.

In the inter mode, the motion predictor 111 may obtain a motion vector by searching for a region that best matches an input block in the reference image stored in the reference picture buffer 190 in the motion prediction process. The motion compensator 112 may generate a prediction block by performing motion compensation using the motion vector.

The subtractor 125 may generate a residual block by the difference between the input block and the generated prediction block. The transformer 130 may perform transform on the residual block and output a transform coefficient. The quantization unit 140 may output the quantized coefficient by quantizing the input transform coefficient according to the quantization parameter.

The entropy encoder 150 may output a bit stream by performing entropy encoding based on the values calculated by the quantizer 140 or the encoding parameter values calculated in the encoding process.

When entropy coding is applied, a small number of bits are allocated to a symbol having a high probability of occurrence and a large number of bits are allocated to a symbol having a low probability of occurrence, thereby representing bits for encoding symbols. The size of the heat can be reduced. Therefore, compression performance of image encoding may be increased through entropy encoding. The entropy encoder 150 may use an encoding method such as exponential golomb, context-adaptive variable length coding (CAVLC), or context-adaptive binary arithmetic coding (CABAC) for entropy encoding.

Since the image encoding apparatus according to the embodiment of FIG. 1 performs inter prediction encoding, that is, inter prediction encoding, the currently encoded image needs to be decoded and stored to be used as a reference image. Accordingly, the quantized coefficients are inversely quantized by the inverse quantizer 160 and inversely transformed by the inverse transformer 170. The inverse quantized and inverse transformed coefficients are added to the prediction block through the adder 175 and a reconstructed block is generated.

The reconstruction block passes through the filter unit 180, and the filter unit 180 applies at least one or more of a deblocking filter, a sample adaptive offset (SAO), and an adaptive loop filter (ALF) to the reconstruction block or the reconstruction picture. can do. The filter unit 180 may be referred to as an adaptive in-loop filter. The deblocking filter can remove block distortion generated at the boundary between blocks. SAO may add an appropriate offset value to the pixel value to compensate for coding errors. The ALF may perform filtering based on a value obtained by comparing the reconstructed image with the original image. The reconstructed block that has passed through the filter unit 180 may be stored in the reference picture buffer 190.

2 is a block diagram illustrating a configuration of an image decoding apparatus according to an embodiment of the present invention.

2, the image decoding apparatus 200 may include an entropy decoder 210, an inverse quantizer 220, an inverse transformer 230, an intra predictor 240, a motion compensator 250, and an adder. 255, a filter unit 260, and a reference picture buffer 270.

The image decoding apparatus 200 may receive a bitstream output from the encoder and perform decoding in an intra mode or an inter mode, and output a reconstructed image, that is, a reconstructed image. In the intra mode, the switch may be switched to intra, and in the inter mode, the switch may be switched to inter. The image decoding apparatus 200 may obtain a reconstructed residual block from the input bitstream, generate a prediction block, and then add the reconstructed residual block and the prediction block to generate a reconstructed block, that is, a reconstruction block. .

The entropy decoder 210 may entropy decode the input bitstream according to a probability distribution to generate symbols including symbols in the form of quantized coefficients. The entropy decoding method is similar to the entropy coding method described above.

When the entropy decoding method is applied, a small number of bits are allocated to a symbol having a high probability of occurrence and a large number of bits are allocated to a symbol having a low probability of occurrence, whereby the size of the bit string for each symbol is increased. Can be reduced. Therefore, the compression performance of the image decoding can be increased through the entropy decoding method.

The quantized coefficients are inversely quantized by the inverse quantizer 220 and inversely transformed by the inverse transformer 230, and as a result of the inverse quantization / inverse transformation of the quantized coefficients, a reconstructed residual block may be generated.

In the intra mode, the intra predictor 240 may generate a predictive block by performing spatial prediction using pixel values of blocks already encoded around the current block. In the inter mode, the motion compensator 250 may generate a predictive block by performing motion compensation using the reference image stored in the motion vector and the reference picture buffer 270.

The reconstructed residual block and the prediction block may be added through the adder 255, and the added block may pass through the filter unit 260. The filter unit 260 may apply at least one or more of the deblocking filter, SAO, and ALF to the reconstructed block or the reconstructed picture. The filter unit 260 may output a reconstructed image, that is, a reconstructed image. The reconstructed picture may be stored in the reference picture buffer 270 and used for inter prediction.

As described above with reference to FIGS. 1 and 2, the encoder and the decoder may perform entropy encoding and entropy decoding, respectively. When entropy encoding / decoding is applied, a small number of bits are allocated to a symbol having a high probability of occurrence, and a large number of bits are allocated to a symbol having a low probability of occurrence, thereby representing a symbol string. The size can be reduced. Methods used for entropy encoding / decoding may include exponential gollum, CAVLC, CABAC, and the like.

For example, the encoder and the decoder may store a table for performing entropy encoding / decoding such as a variable length coding (VLC) table, and the encoder and the decoder may perform entropy encoding / decoding using the stored variable length encoding table. Can be done.

The encoder and the decoder may derive the binarization method of the target symbol and the probability model of the target symbol / bin, and then perform entropy encoding / decoding using the derived binarization method or the probability model. have.

Here, binarization means expressing a symbol value as a binary sequence (bin sequence / string). A bin means the value of each binary number (0 or 1) when the symbol is represented as a column of binary numbers through binarization. The probability model refers to a predicted probability of a symbol / bin to be encoded / decoded, which can be derived through a context information / context model. The context information / context model refers to information for determining the probability of a symbol / bin to be encoded / decoded.

More specifically, the CABAC entropy encoding method binarizes non-binarized symbols into bins and converts the context model using encoding information of neighboring and encoding target blocks or information of symbols / bins encoded in a previous step. The bitstream may be generated by performing an arithmetic encoding of the bin by predicting the occurrence probability of the bin according to the determined context model.

That is, the encoder / decoder can effectively perform entropy encoding / decoding by using encoding / decoding information of neighboring blocks and occurrence probability of bins encoded / decoded in a previous step. In addition, in the encoding step, the encoder may select a context model through encoding information of the neighboring block and update the occurrence probability information of the bin generated according to the selected context model.

When a single encoder and / or a single decoder is used for a UHD image having an HD size or more, a large amount of work required for a single processor may be very high and the image processing speed may be very slow. Accordingly, a method for improving encoding efficiency through parallelization of encoding and / or decoding processes may be provided. To this end, a parallelization method for processes such as motion compensation, image interpolation, discrete cosine transform (DCT), quantization, and the like may be provided. The parallelization method may also be applied to the above-described entropy encoding / decoding process.

Parallel entropy encoding / decoding may be performed using a plurality of entropy encoders and / or a plurality of entropy decoders. Techniques used for parallel entropy encoding / decoding include Probability Interval Partitioning Entropy, which maps and / or converts a variable length bin sequence into a variable length codeword. ) And variable length to variable length (V2V). The PIPE and V2V schemes can be applied to both single entropy encoders / decoders and parallel entropy encoders / decoders.

In the parallel entropy encoding / decoding process, the probability of a bin may be divided according to a quantization interval. Each of the divided probability intervals may be allocated to a bin encoder corresponding to each probability interval on the encoder side, and may be allocated to a bin decoder corresponding to each probability interval on the decoder side. In addition, each of the divided probability intervals may have a representative probability value corresponding to each of the probability intervals. When a PIPE or V2V technique is used for parallel entropy encoding / decoding, the plurality of entropy encoders / decoders are encoded using a Least Probable Bin (LPB) probability interval assigned thereto and a representative LPB probability value corresponding to the probability interval. / Decryption can be performed. Here, LPB may mean a value of a bin that is less than a value of bins of 0 and 1. Hereinafter, in embodiments to be described below, the probability interval may mean an LPB probability interval, and the representative probability may mean a representative LPB probability.

In the encoder, each parallelized entropy encoder codes the bins with a fixed representative probability, and the encoder can send a set of output codewords in the form of a bitstream to the decoder. For example, the encoder may binarize a symbol to be encoded into a bin and predict and / or derive an LPB probability value for each of the bin to be encoded through a context modeler and / or a probability predictor. The encoder may derive a representative LPB probability for each of the probability interval and the probability interval for each bin using the LPB probability value predicted by the context modeler and / or the probability predictor. Each bin may be entropy encoded by a bin encoder to which a corresponding probability interval is assigned. In this case, a bin sequence input to each bin encoder may be converted into a codeword by a codeword table. That is, the bin sequence input to each bin encoder may be output in a codeword form through codeword table mapping. The output codeword may be included in the bitstream and transmitted to the decoder in units of NAL (Network Abstraction Layer).

The bitstream transmitted to the decoder may be parsed by the decoder. In the parsing process, the decoder may map and / or convert a variable length codeword into a variable length empty sequence. At this time, in the comparison process for each bin, overhead may occur.

In addition, the number of entropy encoders included in the encoder may be different from the number of entropy decoders included in the decoder. For example, when the encoded bitstream is transmitted to the decoder and decoded, the number of entropy decoders is entropy. It may be smaller than the number of encoders. In this case, for example, each entropy decoder may parse a bitstream corresponding to one representative probability and then parse the bitstream corresponding to another representative probability. After parsing, each bin is decoded into the value of a meaningful syntax element through de-binarization, so the parsing process described above may cause a delay of the debinarization step, which may result in full entropy. It can greatly affect the decoding speed.

In order to reduce the delay in the inverse binarization step, each entropy encoder included in the encoder can add information about the decodable position in the bitstream to the header of the bitstream, wherein the information about the decodable position is included in the header. The bitstream included in may be transmitted to the decoder. In the present specification, the decodable position may mean a point at which entropy decoding starts and / or a point at which entropy decoding ends in the minimum unit bitstream. In addition, the minimum unit bitstream may indicate a minimum unit in which entropy decoding is performed in a bitstream transmitted from an encoder to the decoder. For example, the minimum unit bitstream corresponds to one codeword. Can be. In this case, each entropy decoder may perform entropy decoding from the midpoint of the bitstream (for example, the start point of the minimum unit bitstream), not the start point of the entire bitstream. The decoder may perform scheduling on the bitstream parsing process in each entropy decoder by using information about a decodable position included in the header of the bitstream. However, even when the information about the decodable position is used as described above, the bin before parsing after the debinarization cannot be recognized as a value of a meaningful syntax element, so a delay may occur in the debinarization step. And the output of the entropy decoder may be delayed.

As described above, when a delay occurs in the parsing process and the inverse binarization process, the parallel entropy encoding / decoding speed may be reduced. Accordingly, there is a need for a parallel entropy encoding / decoding method capable of minimizing delay in parsing and inverse binarization.

3 is a table schematically illustrating one embodiment of a codeword length restriction according to the present invention. 3 illustrates the minimum and maximum lengths of codewords for each of the fixed probability values of LPB, which may be representative LPB probabilities.

When a codeword that is an output of the entropy encoder is generated, the length of the codeword may have a variable value. However, as described above, when a variable length codeword is converted and / or mapped to a variable length empty sequence in the decoder, an overhead may occur in the codeword parsing process because a large amount of computation is required. Thus, a predetermined fixed length codeword may be used, in which case the parsing speed of the decoder can be improved. However, when a codeword of a predetermined fixed length is used, a problem may occur that the coding efficiency is reduced.

Therefore, in order to reduce the overhead incurred in the parsing process, the length of the codeword may be limited to a minimum length or more and a maximum length or less. The minimum length and the maximum length of the codeword may be defined for each representative LPB probability that the parallel entropy encoder has. That is, the minimum length and the maximum length of the codeword may be defined in consideration of the representative LPB probability value of each entropy encoder in the encoder.

For example, the encoder may include a codeword length calculator. The codeword length calculator may derive an optimal codeword length (minimum length and maximum length) in consideration of coding efficiency for each fixed probability value that may be a representative probability. In this case, each entropy encoder included in the encoder may generate a codeword having a length greater than or equal to the minimum length and less than or equal to the maximum length according to the representative probability value corresponding thereto. 3 illustrates an embodiment of the minimum length and the maximum length of the derived codeword.

Referring to FIG. 3, for each fixed probability value, the maximum length and the minimum length of a codeword may be derived. For example, when the fixed probability value is 0.5, the maximum length of the codeword may be 8 and the minimum length of the codeword may be 6. In addition, when the fixed probability value is 0.005, the maximum length of the codeword may be 13 and the minimum length of the codeword may be 1.

In addition, in the embodiment of FIG. 3, as the fixed probability value is higher, the maximum length value of the codeword may be smaller and the minimum length value may be larger. Therefore, the larger the fixed probability value of the LPB, the smaller the variance of the codeword length value, and the smaller the fixed probability value of the LPB, the larger the variance of the codeword length value.

The minimum length and the maximum length of the codeword derived for each fixed probability value are not limited to the above-described embodiment, and may be derived in various forms as the case may be. Hereinafter, the information about the minimum length and the maximum length of the codeword derived from the encoder is referred to as codeword length information.

Codeword length information derived from the encoder may be included in the header of the bitstream and transmitted to the decoder. In this case, the decoder may map or convert the codeword into an empty sequence using the transmitted codeword length information, and the parallel entropy decoding speed may be improved.

4 is a conceptual diagram schematically showing an embodiment of a scheduling method for a bitstream parsing process according to the present invention.

4 illustrates a scheduling method when the number of bin decoders included in the encoder and the number of bin decoders included in the decoder are the same. In the embodiment of FIG. 4, it is assumed that the number of empty encoders and the number of empty decoders are four.

4 shows an embodiment of a bitstream output from the encoder and transmitted to the decoder. Each block constituting the bitstream (blocks separated by a dotted line) may represent a minimum unit bitstream. For example, each minimum unit bitstream may correspond to one codeword. The number indicated in each minimum unit bitstream may indicate the order in which inverse binarization is performed after parsing. In addition, each of 411, 413, 415, and 417 of FIG. 4 may represent a set of all minimum unit bitstreams corresponding to a representative probability of one empty encoder. Hereinafter, a set of all the minimum unit bitstreams corresponding to one representative probability and / or one probability interval is referred to as a representative probability unit bitstream.

420 of FIG. 4 illustrates one embodiment of a method of allocating the bitstream of FIG. 410 to cores of four bin decoders. Referring to 420 of FIG. 4, one representative probability unit bitstream may be allocated to a core of one empty decoder. Here, the core may mean a processor that performs an operation. In this case, each bin decoder may perform entropy decoding on the representative probability unit bitstream allocated thereto.

After parsing, each bin can be decoded into the value of a meaningful syntax element through debinarization. In this case, the parsing process of converting the bitstream into the empty sequence may be performed independently of the inverse binarization process. For example, the order in which the bitstream is converted into an empty sequence and the order in which inverse binarization is performed may be determined separately. Therefore, in 420 of FIG. 4, parsing of the bin required in the debinarization step may be delayed, thereby causing delay in the debinarization process. That is, the parsing process according to the embodiment of 420 of FIG. 4 may cause a delay of the inverse binarization step, which may cause a delay of the entire entropy decoder. Accordingly, an entropy decoding method can be provided that can reduce the delay caused in the inverse binarization process.

430 of FIG. 4 shows another embodiment of a bitstream output from an encoder and transmitted to a decoder. Each block constituting the bitstream (blocks separated by a dotted line) may represent a minimum unit bitstream. For example, each minimum unit bitstream may correspond to one codeword. The number indicated in each minimum unit bitstream may indicate the order in which inverse binarization is performed after parsing. 433, 435, 437, and 439 of FIG. 4 may represent a representative probability unit bitstream.

In addition, the bitstream header 431 of 430 of FIG. 4 may include scheduling information for scheduling of the bitstream parsing process. Here, scheduling may refer to a process of reordering bitstreams in consideration of a debinarization order for syntax elements of a symbol. Once scheduling is performed, the reordered bitstream may be assigned to the empty decoder. Specific embodiments of the scheduling information will be described later.

440 of FIG. 4 illustrates an embodiment of a method of allocating the bitstream of FIG. 430 to cores of four bin decoders.

Referring to 440 of FIG. 4, one representative probability unit bitstream may be classified and / or divided into a plurality of partial bitstreams. Here, the partial bitstream may represent an entropy decoding unit determined by the encoder for reordering the bitstream, and one partial bitstream may include one or a plurality of minimum unit bitstreams. When the representative probability unit bitstream is divided into partial bitstreams, each entropy decoder is not at the beginning of the entire representative probability unit bitstream, but at the midpoint of the bitstream (eg, at the beginning of the partial bitstream). You can start parsing.

The decoder may reorder the partial bitstreams to determine the bitstreams allocated to each bin decoder, so that the bins that are needed first during the inverse binarization process may be parsed first in consideration of the inverse binarization order for the syntax elements of the symbol. In this case, the decoder may allocate the bitstream so that the plurality of bin decoders may have similar workloads in consideration of the number of bin decoders.

The point at which the representative probability unit bitstream is divided and / or classified into partial bitstreams may be determined at the encoder, and the priority of parsing for each representative probability unit bitstream and / or each partial bitstream may also be determined at the encoder. have. In the determination process, the encoder may determine the partial bitstream and priority in consideration of the inverse binarization order for the syntax elements of the symbol.

Hereinafter, the information about the point where the representative probability unit bitstream is divided and / or classified into the partial bitstream is referred to as partial bitstream information, and for each representative probability unit bitstream and / or each partial bitstream, Information about the priority of parsing is called priority information. The above-described scheduling information may include the partial bitstream information and the priority information, and may include the additional information when additional information necessary for scheduling exists.

When scheduling information is determined, the entropy encoder may add the determined scheduling information to the header 431 of the bitstream and transmit the scheduling information to the decoder. The decoder may recognize the high priority representative probability unit bitstream and / or the high priority partial bitstream based on the transmitted scheduling information, and perform parsing on a partial bitstream basis. The decoder may perform scheduling for the bitstream parsing process based on the scheduling information, thereby minimizing a delay occurring in the inverse binarization step.

According to the above-described method, smooth job distribution between a plurality of bin decoders included in the decoder can be achieved, and delay in the debinarization step can be minimized. Therefore, the output delay in the parallel entropy decoder can be minimized, and the process of performing entropy decoding can be speeded up.

5 is a conceptual diagram schematically showing another embodiment of a scheduling method for a bitstream parsing process according to the present invention.

5 illustrates a scheduling method when the number of bin decoders included in the decoder is smaller than the number of bin encoders included in the encoder. In the embodiment of FIG. 5, it is assumed that the number of empty encoders is four and the number of empty decoders is three.

510 of FIG. 5 shows an embodiment of a bitstream output from an encoder and transmitted to a decoder. Each block constituting the bitstream (blocks separated by a dotted line) may represent a minimum unit bitstream. For example, each minimum unit bitstream may correspond to one codeword. The number indicated in each minimum unit bitstream may indicate the order in which inverse binarization is performed after parsing. In addition, each of 511, 513, 515, and 517 of FIG. 5 may represent a representative probability unit bitstream. In 510 of FIG. 5, there may be a total of four representative probability unit bitstreams.

520 of FIG. 5 illustrates an embodiment of a method of allocating the bitstream of FIG. 510 to cores of three bin decoders. Referring to 520 of FIG. 5, one representative probability unit bitstream may be allocated to a core of one empty decoder. In this case, each bin decoder may perform entropy decoding on the representative probability unit bitstream allocated thereto.

In 520 of FIG. 5, since the number of empty decoders is three, when parsing starts, one representative probability unit bitstream may be allocated to each of the three empty decoders. In this case, the representative probability unit bitstream that is not allocated to the empty decoder may be parsed after at least one of the three representative probability unit bitstreams allocated to the empty decoder is parsed. In this case, bins with a higher binarization priority may be parsed later than bins with a lower binarization priority, which may cause a delay in the binarization process. That is, the parsing process according to the embodiment of 520 of FIG. 5 may cause a delay of the inverse binarization step, which may cause a delay of the entire entropy decoder. Accordingly, an entropy decoding method can be provided that can reduce the delay caused in the inverse binarization process.

530 of FIG. 5 shows another embodiment of a bitstream output from an encoder and transmitted to a decoder. Each block constituting the bitstream (blocks separated by a dotted line) may represent a minimum unit bitstream. For example, each minimum unit bitstream may correspond to one codeword. The number indicated in each minimum unit bitstream may indicate the order in which inverse binarization is performed after parsing. In addition, each of 533, 535, 537, and 539 of FIG. 5 may represent a representative probability unit bitstream.

In addition, the bitstream header 531 of 530 of FIG. 5 may include scheduling information for scheduling of the bitstream parsing process. Here, the scheduling information may include partial bitstream information and priority information, and may include the additional information when additional information necessary for scheduling exists. When scheduling is performed, the reordered bitstream may be allocated to three empty decoders.

540 of FIG. 5 illustrates an embodiment of a method of allocating the bitstream of FIG. 530 to cores of three bin decoders.

Referring to 540 of FIG. 5, one representative probability unit bitstream may be classified and / or divided into a plurality of partial bitstreams. Here, one partial bitstream may include one or a plurality of minimum unit bitstreams. If the representative probability unit bitstream is divided into partial bitstreams, each entropy decoder may be located at an intermediate point of the bitstream (eg, at the beginning of the partial bitstream) rather than at the beginning of the entire representative probability unit bitstream. You can start parsing.

The decoder may reorder the partial bitstreams to determine the bitstreams allocated to each bin decoder, so that the bins that are needed first during the inverse binarization process may be parsed first in consideration of the inverse binarization order for the syntax elements of the symbol. In this case, the decoder may allocate the bitstream so that the plurality of bin decoders may have similar workloads in consideration of the number of bin decoders.

Scheduling information (partial bitstream information and priority information) necessary for performing the above-described scheduling process may be determined in the encoder. In the determination process, the encoder may determine the partial bitstream and priority in consideration of the inverse binarization order for the syntax elements of the symbol. When the scheduling information is determined, the entropy encoder may add the determined scheduling information to the header 531 of the bitstream and transmit the scheduling information to the decoder.

The decoder may recognize a high priority representative probability unit bitstream and / or a high priority partial bitstream based on the transmitted scheduling information, and perform parsing on a partial bitstream basis. The decoder may perform scheduling for the bitstream parsing process based on the scheduling information, thereby minimizing a delay occurring in the inverse binarization step.

According to the above-described method, smooth job distribution between a plurality of bin decoders included in the decoder can be achieved, and delay in the debinarization step can be minimized. Therefore, the output delay in the parallel entropy decoder can be minimized, and the process of performing entropy decoding can be speeded up.

6 is a block diagram schematically illustrating an embodiment of a parallel entropy encoding apparatus according to the present invention.

The parallel entropy encoding apparatus according to the embodiment of FIG. 6 includes a binarizer 610, a probability estimator 620, a probability quantizer 630, a codeword length calculator 640, It may include a mapping table calculator 650, an empty sequence calculator 660, a bin encoder 670, a buffer 680, and a bitstream determiner 690. The parallel entropy encoding apparatus according to the embodiment of FIG. 6 may include a plurality of empty encoders 670 to perform parallel entropy encoding.

Referring to FIG. 6, the binarizer 610 may binarize each syntax element obtained in the encoding step. At this time, each syntax element may be converted into an empty sequence by the binarization process. Here, the empty sequence may be composed of a combination of 0's and 1's. The probability predictor 620 may predict the LPB probability for each of the bins obtained by the binarization process.

The probability quantizer 630 may divide the probability intervals and perform LPB probability quantization on each of the bins in order to increase coding efficiency. That is, the probability quantizer 630 may determine which probability interval each bin corresponds to using the LPB probability of the predicted bin. The probability quantizer 630 may determine, for each bin, a bin encoder 670 used for entropy encoding among a plurality of bin encoders, according to a probability section corresponding to each bin.

The codeword length calculator 640 may derive and / or obtain an optimal codeword length (minimum length and maximum length) in consideration of coding efficiency for each representative probability value. In this case, the codeword length calculator 640 may determine the minimum length and the maximum length of the codeword based on the characteristics of the representative probability values in order to maintain coding efficiency.

In one embodiment, the information about the minimum length and the maximum length of the derived codeword (codeword length information) may be included in the header of the bitstream and transmitted to the decoder. In this case, the decoder may determine the minimum length and the maximum length of the codeword parsed by each bin decoder using the codeword length information included in the header.

In another embodiment, the maximum length and minimum length of the codeword for each representative probability may be derived using the same algorithm in the encoder and the decoder. In this case, the maximum length and the minimum length of the codeword may be derived based on a representative probability value of each bin encoder.

According to the above-described embodiment, the codeword length calculator 640 limits the length of the codeword to a certain range, thereby improving the parsing speed for the bitstream.

The mapping table calculator 650 may select and / or generate a mapping table by using codeword length information generated by the codeword length calculator 640. In this case, the mapping table may be used in a process of converting and / or mapping an empty sequence into a codeword. Thus, the mapping table may be referred to as a codeword-bin mapping table.

In one example, the minimum length and maximum length of the codeword may be some fixed value. In this case, the encoder and the decoder may store the same mapping table for each codeword length value within the above range. When the minimum length and the maximum length of the codeword are the same, one fixed length mapping table may be stored.

The encoder can select and use one of the stored mapping tables. In addition, the encoder may add mapping table selection information indicating whether a mapping table is used among the stored mapping tables to the decoder of the bitstream and transmit the mapping table selection information to the decoder. In this case, the decoder may select one mapping table from among the stored plurality of mapping tables using the transmitted mapping table selection information.

As another example, the mapping table may be generated using the same generation algorithm in the encoder and the decoder. In this case, the encoder and the decoder may generate a mapping table corresponding to each codeword length within a codeword length range determined by a minimum length and a maximum length.

In addition to the generation algorithm, if there is additional information for generating a mapping table, the encoder may add the additional information to the header of the bitstream and transmit the additional information to the decoder. Here, the additional information may include, for example, information regarding a maximum length and a minimum length of a codeword. In this case, the decoder may generate a mapping table using information on the maximum length and the minimum length of the transmitted codeword.

In the above-described case, the encoder may select and use one of the generated mapping tables. In addition, the encoder may add mapping table selection information indicating whether a mapping table is used among the generated mapping tables to the decoder of the bitstream and transmit the mapping table selection information to the decoder. In this case, the decoder may select one mapping table from among the plurality of mapping tables using the transmitted mapping table selection information.

The bin sequence calculator 660 may derive information about the binarization / debinarization order for the syntax elements of the symbol and information about the probability of each bin. The empty sequence calculator 660 may derive the scheduling information for the bitstream parsing process based on the derived information. Here, scheduling may refer to a process of reordering bitstreams in consideration of a debinarization order for syntax elements of a symbol.

As described above with reference to FIG. 4, the scheduling information may include partial bitstream information and priority information, and may include the additional information when additional information necessary for scheduling exists. Here, the partial bitstream information indicates information regarding a point at which the representative probability unit bitstream is divided into partial bitstreams, and the priority information indicates the priority of parsing for each representative probability bitstream and / or each partial bitstream. It can represent information about.

As described above, the number of empty encoders included in the encoder may be different from the number of empty decoders included in the decoder. For example, the number of empty decoders may be smaller than the number of empty decoders. In this case, bins with a higher binarization priority may be parsed later than bins with a lower binarization priority, which may cause a delay in the binarization process. At this time, the scheduling information derived from the empty sequence calculator 660 is used in the entropy decoding process, thereby minimizing the delay incurred in the inverse binarization process. To add and / or include the derived scheduling information to the header of the bitstream, the empty sequence calculator 660 can send the scheduling information to the bitstream determiner 690.

When the information derived from the codeword length calculator 640 and the empty sequence calculator 660 described above is used for entropy decoding, the output delay of the parallel entropy decoder can be minimized. Therefore, the process of performing entropy decoding can be speeded up.

The bin encoder 670 may perform entropy encoding for each bin by using the mapping table derived by the mapping table calculator 650. Each bin encoder 670 may convert binarized bins into codewords using a mapping table corresponding to a representative probability. The codeword output from the empty encoder 670 may be stored in the buffer 680.

Codewords stored in the buffer may be transmitted or stored to the decoder through the bitstream determiner 690. The bitstream determiner 690 may add header information to the header of the bitstream. The header information may include, for example, codeword length information, mapping table selection information, information added for generating a mapping table, and scheduling information. The bitstream determiner 690 may transmit the generated bitstream to the decoder. The decoder can minimize the decoding delay and improve the output speed by limiting the codeword length or reordering the bitstream using the header information included in the transmitted bitstream.

7 is a block diagram schematically illustrating an embodiment of a parallel entropy decoding apparatus according to the present invention.

The parallel entropy decoding apparatus according to the embodiment of FIG. 7 includes a codeword length calculator 710, a mapping table calculator 720, a bitstream determiner 730, an empty decoder 740, a buffer 750, and a probability quantizer ( 760, probability predictor 770, and inverse binarizer 780. The parallel entropy decoding apparatus according to the embodiment of FIG. 7 may include a plurality of bin decoders 740 to perform parallel entropy decoding.

Referring to FIG. 7, the codeword length calculator 710 may derive codeword length information from a header of a bitstream transmitted by an encoder.

As described above, due to the variable length of the codeword, overhead may occur in the bitstream parsing process, thereby reducing the parsing speed. In order to solve such a problem, in the encoder, the minimum length and the maximum length of the codeword may be determined for each representative LPB probability of the parallel entropy encoder.

In one embodiment, the information about the minimum length and the maximum length of the derived codeword (codeword length information) may be included in the header of the bitstream and transmitted to the decoder. In this case, the decoder may determine the minimum length and the maximum length of the codeword parsed by each bin decoder using the codeword length information included in the header.

In another embodiment, the maximum length and minimum length of the codeword for each representative probability may be derived using the same algorithm in the encoder and the decoder. In this case, the maximum length and the minimum length of the codeword may be derived based on a representative probability value of each bin encoder.

According to the above-described embodiment, the codeword length calculator 710 limits the length of the codeword to a certain range, thereby improving the parsing speed for the bitstream.

The mapping table calculator 720 may select and / or generate a mapping table using the codeword length information derived by the codeword length calculator 710. In this case, the mapping table may be selected and / or generated based on a representative probability of each bin decoder 740.

In one example, the minimum length and maximum length of the codeword may be some fixed value. In this case, the encoder and the decoder may store the same mapping table for each codeword length value within the above range. When the minimum length and the maximum length of the codeword are the same, one fixed length mapping table may be stored.

The encoder can select and use one of the stored mapping tables. In addition, the encoder may add mapping table selection information indicating whether a mapping table is used among the stored mapping tables to the decoder of the bitstream and transmit the mapping table selection information to the decoder. In this case, the decoder may select one mapping table from among the stored plurality of mapping tables using the transmitted mapping table selection information.

As another example, the mapping table may be generated using the same generation algorithm in the encoder and the decoder. In this case, the encoder and the decoder may generate a mapping table corresponding to each codeword length within a codeword length range determined by a minimum length and a maximum length.

In addition to the generation algorithm, if there is additional information for generating a mapping table, the encoder may add the additional information to the header of the bitstream and transmit the additional information to the decoder. Here, the additional information may include, for example, information regarding the maximum length and the minimum length of the codeword, and the maximum length and the minimum length of the codeword may be determined for each representative LPB probability of the parallel entropy encoder. . In this case, the decoder may generate a mapping table using information on the maximum length and the minimum length of the transmitted codeword.

In the above-described case, the encoder may select and use one of the generated mapping tables. In addition, the encoder may add mapping table selection information indicating whether a mapping table is used among the generated mapping tables to the decoder of the bitstream and transmit the mapping table selection information to the decoder. In this case, the decoder may select one mapping table from among the plurality of mapping tables using the transmitted mapping table selection information.

In the above embodiment, the mapping table is determined using codeword length information transmitted from the encoder. Accordingly, the length of the codeword for each representative probability may be limited, the delay of the parsing process may be minimized, and the output speed of the empty decoder 740 may be improved.

The bitstream determiner 730 may decode the header information included in the header of the bitstream transmitted from the encoder. The header information may include, for example, scheduling information (partial bitstream information and / or priority information). The bitstream determiner 730 may classify each representative probability unit bitstream into a partial bitstream using the scheduling information, and may perform scheduling on the bitstream input to the empty decoder 740. In this case, the decoder may allocate the bitstream to the bin decoder so that the plurality of bin decoders may have similar workloads in consideration of the number of bin decoders. Since a specific embodiment of the scheduling method has been described above with reference to FIGS. 4 and 5, it will be omitted.

The bitstream determiner 730 may minimize the input delay in the inverse binarization step by performing scheduling on the bitstream using scheduling information transmitted from the encoder.

When the information derived from the codeword length calculator 710 and the bitstream determiner 730 described above is used for entropy decoding, the output speed of the parallel entropy decoder may be improved. Therefore, the process of performing entropy decoding can be speeded up.

Each bin decoder 740 may convert a codeword into an empty sequence using a mapping table corresponding to the scheduled bitstream and the representative probability. In this case, the bitstream input to each bin decoder 740 may be determined according to a representative LPB probability.

The bins output from the bin decoder 740 may be input to the probability predictor 770 through the buffer 750 and the probability quantizer 760 in order. The probability predictor 770 may derive the LPB probability for the bin to be input. The empty sequence may be decoded into a value of a meaningful syntax element through the depreciator 780 after passing through the probability predictor 770.

8 is a block diagram schematically showing another embodiment of a parallel entropy encoding apparatus according to the present invention.

The parallel entropy encoding apparatus according to the embodiment of FIG. 8 includes a binarizer 810, a probability predictor 820, a probability quantizer 830, a codeword length calculator 840, a mapping table calculator 850, and a bin. It may include an encoder 860 and a buffer 870. The parallel entropy encoding apparatus according to the embodiment of FIG. 8 may include a plurality of empty encoders 860 to perform parallel entropy encoding.

As described above, the encoder may limit the length of the codeword in order to reduce the overhead incurred in the parsing process to improve the output speed of the decoder. Since a coding efficiency can be reduced when a predetermined fixed length codeword is used, the encoder can determine the minimum length and the maximum length of the codeword for each of the empty encoder and / or the empty decoder.

Referring to FIG. 8, the codeword length calculator 840 may derive and / or obtain an optimal codeword length (minimum length and maximum length) for each representative probability value of an empty encoder in consideration of coding efficiency. have. In this case, the codeword length calculator 840 may determine the minimum length and the maximum length of the codeword based on the characteristics of the representative probability in order to maintain coding efficiency.

In the embodiment of FIG. 8, the operation of the other components except for the above-described configuration is the same as in the embodiment of FIG. 6, and thus will be omitted.

9 is a block diagram schematically showing another embodiment of a parallel entropy decoding apparatus according to the present invention.

The entropy decoding apparatus according to the embodiment of FIG. 9 includes a codeword length calculator 910, a mapping table calculator 920, a bin decoder 930, a buffer 940, a probability quantizer 950, and a probability predictor. 960 and debinarizer 970. The parallel entropy decoding apparatus according to the embodiment of FIG. 9 may include a plurality of bin decoders 930 to perform parallel entropy decoding. The parallel entropy decoding apparatus according to the embodiment of FIG. 9 may operate similarly to the parallel entropy decoding apparatus according to the embodiment of FIG. 7.

The bitstream transmitted from the encoder may be allocated to the plurality of empty decoders 930 included in the decoder. At this time, the decoder may determine the bitstream allocated to each empty decoder using the header information included in the header of the transmitted bitstream. In this case, due to the variable length of the codeword, overhead may occur in the parsing process of the bitstream, thereby reducing the parsing speed. In order to solve this problem, the codeword length calculator 910 may determine the minimum length and the maximum length of the codeword for each representative LPB probability of the parallel entropy encoder.

In one embodiment, information about the minimum length and the maximum length of the codeword derived from the encoder (codeword length information) may be included in the header of the bitstream and transmitted to the decoder. In this case, the codeword length calculator 910 may determine the minimum length and the maximum length of the codeword parsed by each bin decoder using the codeword length information included in the header.

In another embodiment, the maximum length and minimum length of the codeword for each representative probability may be derived using the same algorithm in the encoder and the decoder. In this case, the maximum length and the minimum length of the codeword may be derived based on a representative probability value of each bin encoder.

The mapping table calculator 920 uses information on the representative probability of each bin decoder 930 and information on the minimum length and maximum length of the codeword determined by the codeword length calculator 910. The mapping table can be determined. Since a specific embodiment of the mapping table determination method has been described above with reference to FIG. 7, it will be omitted.

By the codeword length calculator 910 and the mapping table calculator 920 described above, the output speed of the parallel entropy decoder can be improved. Therefore, the process of performing entropy decoding can be speeded up.

In the embodiment of Fig. 9, the operation of the other configurations except for the above-described configuration is the same as in the embodiment of Fig. 7, and thus will be omitted.

10 is a block diagram schematically showing another embodiment of a parallel entropy encoding apparatus according to the present invention.

The parallel entropy encoding apparatus according to the embodiment of FIG. 10 includes a binarizer 1010, a probability predictor 1020, a probability quantizer 1030, an empty sequence calculator 1040, a bin encoder 1050, and a buffer ( 1060 and bitstream determiner 1070. The parallel entropy encoding apparatus according to the embodiment of FIG. 10 may include a plurality of empty encoders 1050 to perform parallel entropy encoding.

Referring to FIG. 10, each syntax element obtained in the encoding process of the binarizer 1010 may be binarized. At this time, each syntax element may be converted into an empty sequence by the binarization process. The probability predictor 1020 may calculate a probability of occurrence for each of the bins obtained by the binarization process. In addition, the probability quantizer 1030 may divide the probability intervals and perform LPB probability quantization on each of the bins.

The empty sequence calculator 1040 may derive scheduling information about the bitstream parsing process. As described above with reference to FIG. 4, the scheduling information may include partial bitstream information and priority information, and may include the additional information when additional information necessary for scheduling exists. Here, the partial bitstream information indicates information regarding a point at which the representative probability unit bitstream is divided into partial bitstreams, and the priority information indicates the priority of parsing for each representative probability bitstream and / or each partial bitstream. It can represent information about.

As described above, in the parsing process, a bin having a high debinarization priority may be parsed later than a bin having a low debinarization priority, and in this case, a delay may occur in the debinarization process. At this time, since the scheduling information derived from the empty sequence calculator 1040 is used in the entropy decoding process, the delay incurred in the inverse binarization process can be minimized. To add and / or include the derived scheduling information to the header of the bitstream, the empty sequence calculator 1040 may send the scheduling information to the bitstream determiner 1070. The bitstream determiner 1070 may add the scheduling information to the header of the bitstream, and the bitstream may be transmitted to the decoder.

By the empty sequence calculator 1040 described above, the output speed of the parallel entropy decoder can be improved. Therefore, the process of performing entropy decoding can be speeded up.

In the embodiment of FIG. 10, the operation of the other components except for the above-described configuration is the same as in the embodiment of FIG. 6, and thus will be omitted.

11 is a block diagram schematically showing another embodiment of a parallel entropy decoding apparatus according to the present invention.

The entropy decoding apparatus according to the embodiment of FIG. 11 includes a bitstream determiner 1110, a bin decoder 1120, a buffer 1130, a probability quantizer 1140, a probability predictor 1150, and an inverse binarizer ( 1160). The parallel entropy decoding apparatus according to the embodiment of FIG. 11 may include a plurality of bin decoders 930 to perform parallel entropy decoding. The parallel entropy decoding apparatus according to the embodiment of FIG. 11 may operate similarly to the parallel entropy decoding apparatus according to the embodiment of FIG. 7.

The bitstream determiner 1110 may decode the header information included in the header of the bitstream transmitted from the encoder. The header information may include, for example, scheduling information (partial bitstream information, priority information, and / or additional information required for bitstream reordering). The bitstream determiner 1110 may classify each representative probability unit bitstream into a partial bitstream by using the scheduling information, and may perform scheduling on the bitstream input to the empty decoder 1120. In this case, the decoder may determine a bitstream allocated to each bin decoder so that the plurality of bin decoders may have a similar workload in consideration of the number of bin decoders.

The bitstream determiner 1110 may minimize the input delay in the inverse binarization step by performing scheduling on the bitstream using the scheduling information transmitted from the encoder.

*

In the embodiment of Fig. 11, the operation of the configuration except for the above-described configuration is the same as in the embodiment of Fig. 7, and thus will be omitted.

12 is a flowchart schematically illustrating an embodiment of a parallel entropy encoding method according to the present invention.

Referring to FIG. 12, the encoder may convert syntax elements into an empty sequence using a predetermined binarization method (S1210). Here, the bin sequence may be composed of a combination of 0's and 1's, and each 0 or 1 may be called a bin.

The encoder may determine the LPB probability for each binarized bin (S1220), and divide the probability interval of the bin according to the quantization interval (S1230).

In addition, the encoder may derive the scheduling information in order to minimize the input delay in the debinarization step of the decoder (S1240). As described above, the scheduling information may include partial bitstream information and priority information, and may include the additional information when additional information necessary for scheduling exists. The derived scheduling information may be included in the header of the bitstream and transmitted to the decoder.

As described above, the probability intervals of bins may be divided according to quantization intervals, and each of the divided probability intervals may be allocated to a bin encoder corresponding to each probability interval. The encoder may determine a bin encoder used for entropy encoding of each bin through quantization according to a probability section corresponding to each bin (S1250).

The encoder can derive codeword length information for each representative LPB probability of the empty encoder by a predetermined algorithm (S1260). Here, the codeword length information may be, for example, information regarding a minimum length and a maximum length of a codeword. When codeword length information is derived, the encoder may determine a mapping table using the derived codeword length information. Here, the mapping table may mean a table used in a process of converting and / or mapping an empty sequence into a codeword.

The encoder may perform bin encoding on each bin by using the determined mapping table (S1270). The encoder may convert the binarized bins into codewords by performing bin encoding. The output codeword may be included in the bitstream and transmitted to the decoder in units of NAL.

13 is a flowchart schematically showing an embodiment of a parallel entropy decoding method according to the present invention.

Referring to FIG. 13, the decoder may receive a bitstream from an encoder (S1310). The header of the received bitstream may include header information. Here, the header information may include, for example, codeword length information, mapping table selection information, information added to generate a mapping table, and scheduling information.

The decoder may derive scheduling information from the header of the bitstream and perform scheduling on the bitstream using the derived scheduling information in order to minimize the input delay in the inverse binarization step (S1320). Here, scheduling may refer to a process of reordering bitstreams in consideration of a debinarization order for syntax elements of a symbol. In this case, the decoder may allocate the bitstream to the bin decoder so that the plurality of bin decoders may have similar workloads in consideration of the number of bin decoders.

Referring back to FIG. 13, the decoder may derive codeword length information from the header of the bitstream transmitted from the encoder (S1330). Here, the codeword length information may mean information on the minimum length and the maximum length of the codeword, corresponding to the representative LPB probability of each bin decoder. When codeword length information is derived, the decoder may determine a mapping table using the derived codeword length information.

The decoder may perform bin decoding on each bin by using the determined mapping table (S1340). In this case, the plurality of bin decoders included in the decoder may decode bins corresponding to probability intervals allocated thereto. The decoder may convert a codeword into a bin by performing bin decoding.

The decoder may determine the LPB probability for each bin output from the bin decoder after the probability quantization step (S1350). In addition, the decoder may perform debinarization on each bin output from the bin decoder to derive a value of a meaningful syntax element (S1360).

In the above-described embodiments, the methods are described based on a flowchart as a series of steps or blocks, but the present invention is not limited to the order of steps, and certain steps may occur in a different order or at the same time than the other steps described above. Can be. Also, one of ordinary skill in the art appreciates that the steps shown in the flowcharts are not exclusive, that other steps may be included, or that one or more steps in the flowcharts may be deleted without affecting the scope of the present invention. I can understand.

The above-described embodiments include examples of various aspects. Although not all possible combinations may be described to represent the various aspects, one of ordinary skill in the art will recognize that other combinations are possible. Accordingly, it is intended that the present invention cover all other replacements, modifications and variations that fall within the scope of the following claims.

Claims (2)

Encoding the image to obtain a syntax element;
Determining decodable position information; And
Generating a bitstream by entropy encoding the syntax element and the decodable position information;
The bitstream includes a plurality of partial bitstreams,
The number of decodable position information is used to determine the number of the plurality of partial bitstreams, and the decodable position information of each of the plurality of partial bitstreams is used to determine an end point of each of the plurality of partial bitstreams. ,
The bitstream includes a header area and a data area,
The decodable position information is included in a header region of the bitstream,
And obtaining a syntax element by encoding the image comprises obtaining a syntax element using at least one prediction method of an intra mode and an inter mode.
A computer-readable recording medium storing a bitstream received by an image decoding apparatus and used to recover a current block included in an image.
The bitstream includes syntax elements and decodable position information,
The bitstream includes a plurality of partial bitstreams,
The number of decodable position information is used to determine the number of the plurality of partial bitstreams, and the decodable position information of each of the plurality of partial bitstreams is used to determine an end point of each of the plurality of partial bitstreams. ,
The bitstream includes a header area and a data area,
The decodable position information is included in a header region of the bitstream,
And the syntax element is obtained using at least one prediction method of an intra mode and an inter mode.

Images