KR20060083100A - Method and apparatus for adaptive entropy encoding and decoding for scalable video coding - Google Patents

Abstract

Disclosed are a method and apparatus for adaptive entropy encoding and decoding using various context models. According to the present invention, the hierarchical entropy encoding method includes (a) determining a context by referring to encoding elements of the same layer and lower layer as the layer of the block to which the encoding element to be encoded, or by referring to only the encoding elements of the lower layer. ; And (b) entropy encoding the encoding element using the determined context. As a result, entropy encoding may be performed using various contexts including hierarchical correlation in the stretching video encoding, thereby improving encoding efficiency.

Description

Adaptive entropy encoding and decoding method for flexible coding and apparatus therefor {Method and apparatus for adaptive entropy encoding and decoding for scalable video coding}

1 is a reference diagram for explaining an example of encoding a video by applying a flexible video encoding technique.

2 is a block diagram of a CABAC encoder;

3A to 3B are reference diagrams for explaining the determination of context;

4 is a configuration diagram of a CABAC decoder;

5 is a configuration diagram of an entropy encoder that encodes using mapping information between a symbol and a codeword;

6 is a configuration diagram of an entropy decoder that decodes using mapping information between a symbol and a codeword;

7A to 7B are reference diagrams for describing an operation of a context determination unit according to the present invention;

8A to 8G are diagrams illustrating an embodiment of an adaptive entropy encoder using a context determiner according to the present invention;

9A to 9G are diagrams illustrating an embodiment of an adaptive entropy decoder using a context determiner according to the present invention;

10A to 10C are flowcharts of an entropy encoding method according to an embodiment of the present invention;

11A to 11C are flowcharts of an entropy decoding method according to an embodiment of the present invention;

12 is a block diagram illustrating an example of a video encoding apparatus having an entropy encoder of the present invention;

13 is a block diagram illustrating an example of a video decoding apparatus having an entropy decoder of the present invention;

The present invention relates to a method and apparatus for adaptive entropy encoding and decoding for scalable coding, and more particularly, to an method and apparatus for adaptive entropy encoding and decoding using various context models. .

Scalable video coding refers to video encoding for appropriately adjusting and transmitting the amount of data to be transmitted according to various communication networks and terminal conditions, and adaptively processing video corresponding to various transmission environments. Is a necessary skill. At present, with the development of mobile communication technology and transmission means and the development of high performance semiconductor and video compression technology, the demand for video service in various transmission environments is rapidly increasing.

However, in order to process video in response to various environmental changes according to channel bandwidth, terminal processing capability, packet loss rate, and user's mobile environment, a video encoding technique developed in consideration of a specific communication environment is adaptively adapted to a transmission environment. Not suitable for performing video encoding. Flexible video encoding technology is one of the intelligent technologies adaptive to such a transmission environment. For example, spatial flexible coding for screen size, temporal flexible coding for frame rate, and Signal to Noise Ratio (SNR) considering image quality Telescopic coding and the like.

Conventional video standards also include this flexible video encoding technique. For example, MPEG-2 flexible coding mainly for video data transmission in Asynchronous Transfer Mode (ATM) networks, SNR of H.263 Annex.O, temporal and spatial flexible coding, and MPEG-4 based micro flexible coding ( Fine Granular Scalable Coding), and the flexible video encoding technology of the MPEG-4 AVC-based standard is currently being standardized. MPEG-4 AVC-based flexible video encoding, which is currently being standardized, aims to simultaneously support flexible video encoding in terms of SNR and temporal / spatial aspects. FIG. 1 encodes video by applying a flexible video encoding technique. It is a reference figure for demonstrating an example to make.

Referring to FIG. 1, it can be seen that flexible coding can be performed in all of SNR, temporal and spatial aspects. The flexible encoding technique is a technique of dividing a video into hierarchical units and encoding the video according to a state of a network, and the enhancement layers are encoded using data of a lower layer.

Referring to the example of video encoding illustrated in FIG. 1 in more detail, when the transmission bit rate of video data is 41 kbps or less, only the base layer 110 is encoded. When the video bit rate is 41 kbps or more and 80 kbps or less, the base layer 110 is used. SNR flexible encoding with improved image quality is performed using the data of ()) to make and encode the first upper layer 120. The base layer 110 and the first upper layer 120 each have a screen size of QCIF (Quarter Common Intermediate Format) and are encoded in units of 15 frames per second.

The second upper layer 130 and the third upper layer 140 have a screen size of CIF and are encoded in units of 30 frames per second. Accordingly, the second upper layer 130 corresponding to a bit rate up to 115 kbps, upsamples the frame of the first upper layer 120 having the QCIF screen size to make the CIF screen size, and predictively encodes the frame to make the intermediate frame. This is accomplished by creating more H-pass frames. The third upper layer 140 is formed at a bit rate of up to 256 kbps, and similar to the first upper layer 120, the SNR flexible encoding having improved image quality with the data of the second upper layer 130, which is the lower layer, is improved. Bi-predictive (B) frames or high-pass (H-frame) frames of each layer can be flexibly encoded in temporal order because only the preceding frames are used as reference images in motion compensation. Referring to FIG. 1, it can be seen that I frames and P frames or L-pass frames precede B frames or H frames in the transmission order. The order of transmission between the same B frames and H frames depends on the assigned index (indicated by superscript in the frame name) as shown in FIG. The lower the index of the same B frames, the higher the index of the H frames, the earlier the transmission order.

For example, in the base layer 110 or the first upper layer 120, B ¹ frames are motion compensated with reference to I frames and P frames, and B ² frames are motion compensated with reference to B ¹ frames. In addition, in the second upper layer 130 and the third upper layer 140, H ³ frames are motion compensated with reference to L ³ frames, and H ² frames are motion compensated with reference to H ³ frames. Accordingly, the frame transmission order in the base layer 110 and the first upper layer 120 is I->P-> B ^1- > B ^2- > B ³ , and the second upper layer 130 and the third higher order are used. The frame transmission order in the layer 140 becomes L ^3- > H ^3- > H ^2- > H ^1- > H ⁰ , and the transmission order of frames having the same index is determined according to the time order in which the frames appear. . Through such temporal stretch coding, spatial stretch coding, and SNR stretch coding, the decoder can decode up to the corresponding layer according to the bit rate of each layer.

The flexible video encoding technology has already been established as a standard from the MPEG-2 standard, and much research has been conducted, but it is currently not commercialized. The biggest reason is that the coding efficiency is bad. That is, compared with non-stretchable video encoding technology, since flexible video encoding performs encoding in a manner that gradually improves the quality of the low-quality base layer, even if video or audio of the same bit rate is severely degraded in some cases Occurs. Hierarchical coding cannot be commercialized unless such coding efficiency problems are overcome.

Accordingly, researches for overcoming the above-mentioned degradation of coding efficiency in hierarchical coding have been actively conducted. For example, when the flexible encoding of the spatial domain is performed, the upsampled lower layer of the image may be used for motion compensation, thereby significantly improving encoding efficiency compared to encoding each layer independently. In other words, since the correlation between the layers is very high, the use of the correlation between the layers makes it possible to obtain high coding efficiency during prediction encoding. However, in the conventional telescopic video encoding technique, such entropy coding is required. Since the encoding is performed according to the same method as the non-stretch video encoding without performing the encoding using the inter-layer correlation, the aforementioned problem of deterioration of encoding efficiency cannot be solved.

Accordingly, the present invention provides a context-based adaptive entropy encoding method using the correlation between inter-layer coding elements and an encoding thereof in order to increase the efficiency of entropy encoding in a flexible video encoder that encodes a plurality of layers. To provide a device.

Another object of the present invention is to provide a content-based adaptive entropy decoding method using correlation between inter-layer coding elements, and a decoding apparatus according thereto.

According to the present invention, (a) determining a context by referring to encoding elements of the same layer and lower layer as the layer of the block to which the encoding element to be encoded, or by referring to only the encoding elements of the lower layer; And (b) entropy-encoding the encoding element using the determined context.

In addition, the technical problem according to the present invention, (a) refer to the encoding element of the same layer as the layer of the block to which the encoding element to be encoded, refer to the encoding element of the lower layer, or encode the same layer and lower layer Converting the encoding element to be encoded into a binary string by referring to all elements; (b) determining a context by referring to encoding elements of the same layer as the layer of the block to which the encoding element to be encoded belongs; And (c) entropy encoding the encoding element using the determined context.

According to the present invention, (a) receiving the entropy-coded data, refer to the encoding element of the same layer and lower layer as the layer of the block to which the entropy-encoded encoding element belongs, or the encoding element of the lower layer. Determining a context with reference only; And (b) entropy decoding the entropy encoded data using the determined context.

In addition, according to the present invention, (a) receiving the entropy-encoded data, determining the context by referring to the encoding element of the same layer as the layer of the block to which the entropy-encoded encoding element belongs; (b) entropy decoding the entropy encoded data using the determined context; And (c) the entropy decoded by referring to an encoding element of the same layer as the layer of the block to which the encoded encoding element belongs, referring to an encoding element of a lower layer, or referring to both encoding elements of the same layer and a lower layer. It is also achieved by an entropy decoding method comprising converting a binary string into a symbol.

Meanwhile, according to another field of the present invention, the technical problem refers to a context in which the context is determined by referring to encoding elements of the same layer and lower layer as the layer of the block to which the encoding element to be encoded belongs, or by referring to only the encoding elements of the lower layer. Decision unit; And an entropy encoding engine that entropy encodes the encoding elements using the determined context.

According to the present invention, the technical problem refers to an encoding element of the same layer as a layer of a block to which an encoding element to be encoded belongs, refers to an encoding element of a lower layer, or both encoding elements of the same layer and a lower layer. A binary string converter configured to convert the encoding element to be encoded into a binary string; A context determination unit that determines a context by referring to encoding elements of the same layer as the layer of the block to which the encoding element to be encoded belongs; And an entropy encoding engine that entropy encodes the encoding element by using the determined context.

In addition, according to the present invention, by receiving the entropy-encoded data, by reference to the encoding element of the same layer and lower layer as the layer of the block to which the entropy-encoded encoding element belongs, or referring to only the encoding element of the lower layer A context determination unit to determine a context; And an entropy decoding engine that entropy decodes the entropy-encoded data using the determined context.

In addition, according to the present invention, the context determination unit for receiving the entropy-encoded data, the context determination unit for determining the context with reference to the encoding element of the same layer as the layer of the block to which the entropy-encoded encoding element belongs; An entropy decoding engine that entropy decodes the entropy coded data using the determined context; And the entropy decoded binary string by referring to an encoding element of the same layer as the layer of the block to which the encoded encoding element belongs, referring to encoding elements of a lower layer, or referring to encoding elements of the same layer and a lower layer. It is also achieved by an entropy decoding apparatus comprising a symbol value converting unit for converting into a symbol.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings.

First, before describing context-based entropy encoding and decoding using the degree of correlation between inter-layer coding elements according to the present invention, a conventional conventional entropy encoding and decoding method will be described. More specifically, context-based Adaptive Binary Arithmetic Coding (CABAC), which has high coding efficiency and is also adopted in the H.264 / MPEG-4 AVC standard, is described in detail. Different methods for adaptive arithmetic coding, Huffman coding and single variable length coding (UVLC) will be described. The present invention is not limited to CABAC but can be applied to both general entropy encoding and decoding methods.

2 is a block diagram of a CABAC encoder.

The CABAC encoder of FIG. 2 receives and encodes each encoding element (Syntax Element, SE), and includes a binary string converter 210, a context determiner 220, and an adaptive arithmetic encoder 250. The encoding element SE refers to data constituting a compressed bitstream, and includes a parameter or various header information and residual information necessary for compressing and expressing an image or interpreting the compressed data. For example, in the H.264 standard, parameters such as entropy_coding_mode_flag indicating which entropy coding scheme is used, slice_type indicating a slice type, mb_type and header information indicating a macroblock type, and coeff_abs_level_minus1 corresponding to a residual information All belong to coding elements.

The binary string converter 210 receives a symbol value of an encoding element SE to be encoded and converts the binary string into a binary string. Binary strings consist of a series of binary values of 0 or 1. Table 1 shows an example of what binary string mb_type, the macroblock type of B slices defined in the H.264 standard, is converted to. . Referring to Table 1, the symbol B_L0_16x16 in the mb_type of the B slice means that a given macro block is motion compensated in units of 16x16 with reference to the L0 (List 0) index and the symbol value is 1. The symbol B_L0_16x16 is converted into a binary string 100 having three binary values through the binary string converter 210.

In Table 1, BinIdx represents a position of each binary value in a binary string. For example, a binary value with BinIdx of 0 means the binary value at the first position in the binary string. The mapping relationship between the symbol value and the binary string is generally defined for each encoding element.

Symbol value of mb_type of B slice (symbol) Binary string 0 (B_Direct_16x16) 0 1 (B_L0_16x16) 1 0 0 2 (B_L1_16x16) 1 0 1 3 (B_Bi_16x16) 1 1 0 0 0 0 4 (B_L0_L0_16x8) 1 1 0 0 0 1 5 (B_L0_L0_8x16) 1 1 0 0 1 0 6 (B_L1_L1_16x8) 1 1 0 0 1 1 ... ... BinIdx 0 1 2 3 4 5 6

Each symbol of a given encoding element SE is converted into a binary string and then entropy coded by the adaptive arithmetic encoder 250 in turn with reference to the context determined by the context determiner 220. The context determiner 220 determines a context for each binary value of 0 and 1 in the binary string. The context refers to the surrounding situation associated with the symbol to be currently encoded. The surrounding situation will be described later. Since the context is used by the probability estimator 230 to estimate probability values for 0 and 1 of each binary value, when the context of the binary value changes, each corresponding probability value also changes. The arithmetic coding engine 240 performs arithmetic coding using the probability value estimated by the probability estimator 230, and updates the probability model of the probability estimator 230 with respect to the encoded 0 or 1. Hereinafter, context determination will be described with reference to FIGS. 3A to 3B.

3A to 3B are reference diagrams for explaining the determination of context.

In FIG. 3A, T 310 is referred to as a context template and consists of a subset of encoding elements encoded before the encoding element to be currently encoded. In some cases, the T 310 may further include a portion of the binary string for the symbol value of the encoding element currently being encoded, or may include only a portion of the binary string in the subset. In addition, the attributes of the previously encoded symbol and the symbol to be currently encoded may be additionally included in the T 310 or may be configured only with the attributes of the symbol. As shown in FIG. 3B, the context template of CABAC includes a block S 330 to which a symbol to be currently encoded belongs, a symbol S ₀ of a left block 340 adjacent to a symbol S ₁ of a neighboring upper block 350, and the like.

For example, in the case of the mb_type symbol in the B slice in Table 1, the context of the binary value where BinIdx is 0 is determined depending on whether the mb_type symbols of the left and upper blocks are Direct. In addition, context is also affected when the left and top blocks are skipped or do not exist. Therefore, the context template required for encoding the first binary value of the mb_type symbol includes the mb_type symbols of the left and upper blocks and skip information corresponding to the property of each block and whether a block exists.

On the other hand, a binary value in which BinIdx is smaller than the binary value currently encoded by the context template may be included. For example, in the case of determining the context of the binary value of BinIdx of 2 in the mb_type symbol of Table 1, the binary value of BinIdx of 1 is used as the context template. Therefore, the context of the binary value of BinIdx of 2 varies depending on whether the binary value of BinIdx of coded 1 is 0 or 1.

The modeling function F receives symbols {S ₀ , S ₁ ,..., S _T-1 } belonging to the context template T 310 for each given symbol value, and sets {C ₀ , which is a context set C 320. C ₁ , ..., C _N-1 } will determine the context. The modeling function corresponding to the binary value of BinIdx of 0 in mb_type of Table 1 is shown in Equation 1 below.

ctxIdx = condTermFlag (A) + condTermFlag (B) + Offset

Here, ctxIdx refers to an index indicating a selected context C, and Offset refers to an initial value of an index indicating a start position of mb_type among tables in which contexts are defined. Where condTermFlag (M) represents the state of block M, if block M does not exist (condition 1), block M is skipped (condition 2), or if the symbol of mb_type of block M is Direct (condition 3) condTermFlag (M) has a value of 0. That is, if any one of condition 1, condition 2, and condition 3 is satisfied, it becomes zero. Otherwise, the value is 1. In the example of mb_type, the modeling function F is expressed as F: T = {S ₀ , S ₁ }-> C = {C ₀ , C ₁ , ..., C _N-1 }. In this case, S ₀ and S ₁ are cases where the left block 340 of the block 330 to be currently encoded and the mb_type of the upper block 350 are used as shown in FIG. 3B, and more specifically S ₀ = condTermFlag (A), S ₁ = condTermFlag (B).

Accordingly, the presence of the block A and the block B, whether or not to skip and mb_type symbols are input to the context template to obtain the context index ctxIdx, and the context having this index is extracted from the context set to determine the corresponding context.

The determined context is used by the adaptive arithmetic encoder 250 to estimate the probability of each binary value. The adaptive arithmetic encoder 250 is composed of a probability estimator 230 and an arithmetic coding engine 240. The probability estimator 230 receives a context corresponding to the binary value generated by the binary string converter 210 and estimates each probability of 0 and 1. The probability value corresponding to the context is determined through the mapping table between the context and the probability. The probability value thus determined is used by the arithmetic coding engine 240 to encode the binary value. The arithmetic coding engine 240 encodes binary values in order, and instructs the probability update unit 230 to update the probability according to the encoded binary value information.

On the other hand, unlike the CABAC encoder, a general arithmetic encoder that arithmically encodes a symbol value without binarization of a symbol has the same structure and function as CABAC encoder except for the binary string converter 210. In this case, the context determiner must determine the context for the symbol value of the symbol, unlike the context determiner for the binary value of the symbol.

4 is a block diagram of a CABAC decoder.

The CABAC decoder extracts symbols from a compressed bitstream consisting of 0s and 1s, which are video data compressed by the encoder. The context determiner 410 determines the context of the binary value to be currently decoded, and determines the context according to each position of the binary value in the binary string to be decoded. The determined context is used by the adaptive arithmetic decoding unit 450 to decode a binary value from the input bit string. The probability estimator 420 receives a context and estimates a probability of 0 and 1 for each binary value. The arithmetic decoding engine 430 transmits the probability value to the probability estimator 420 in order to update the probability of the binary value while sequentially decoding the binary value from the input bit string with the estimated probability value. Binary values sequentially decoded by the adaptive arithmetic decoding unit 450 are converted into symbol data through the symbol value converting unit 440. Mapping between symbols and binary values is performed using the same mapping table used in the binary string converter 210 of the CABAC encoder as shown in Table 1.

On the other hand, unlike the CABAC decoder described above, a general arithmetic decoder in which the arithmetic decoding engine directly generates a symbol value does not require a symbol value converter 440 in the CABAC decoder.

5 is a configuration diagram of an entropy encoder that encodes using mapping information between a symbol and a codeword.

Huffman coding, single variable length coding, and Exp-Golomb coding are used as encoding methods using mapping information between symbols and codewords. Table 2 shows an example of mapping information between symbols and code words.

symbol Context1 (0≤nC <2) Context2 (2≤nC <4) Context 3 (4≤nC <8) Context 4 (8≤nC) Context 5 (nC = -1) 0 One 11 1111 0000 11 01 One 0001 01 0010 11 0011 11 0000 00 0001 11 2 01 10 1110 0000 01 One 3 0000 0111 0001 11 0010 11 0001 00 0001 00 4 0001 00 0011 11 0111 1 0001 01 0001 10 5 001 011 1101 0001 10 001 6 0000 0011 1 0000 111 0010 00 0010 00 0000 11 ≤7 0000 0110 0010 10 0110 0 0010 01 0000 011 8 0000 101 0010 01 0111 0 0010 10 0000 010 9 0001 1 0101 1100 0010 11 0001 01

Referring to FIG. 5, the entropy encoder includes a context determiner 510, an entropy encoding engine 520, and a symbol-codeword mapping information storage unit 530. The entropy encoding engine 520 outputs codewords corresponding to the input symbol values as in the example of Table 2 according to the context determined by the context determination unit 510. In this case, the context determiner 510 determines the context to be applied to the symbol to be encoded based on the symbols belonging to the neighboring blocks of the symbol to be encoded or the previously encoded symbols as the context template. The entropy encoding engine 520 extracts the mapping information corresponding to the predetermined context from the symbol-code mapping information storage unit 530 which stores a plurality of predetermined symbol-code mapping information, and inputs the received symbol. Entropy-encodes. For example, when 0 ≤ nC <2 is satisfied in Table 2, context 1 is selected by the context determiner 510, and accordingly, the entropy encoding engine 520 stores the symbol-code mapping information storage unit 530. The codeword for the input symbol is output using the symbol-codeword mapping information supplied from the C-code.

Where nC represents the average number of non-zero transform coefficients in the upper and left peripheral blocks (S ₀ and S ₁ in FIG. 3B) of the symbol to be encoded. That is, nC means an average value of non-zero transformation coefficients outputted when the number of transformation coefficients of S ₀ and S ₁ blocks is given as an input of the modeling function of FIG. 3A.

6 is a configuration diagram of an entropy decoder that decodes using mapping information between symbols and code words.

Methods of decoding using mapping information between symbols and codewords include Huffman decoding, single variable length decoding, and Exp-Golomb decoding. Referring to FIG. 6, the entropy decoder includes a context determiner 610, an entropy decryption engine 620, and a symbol-code mapping information storage unit 630.

The context determiner 610 determines a context with respect to the entropy coded encoding element. The symbol-codeword mapping information storage unit 630 stores mapping information between a plurality of predefined symbol-codewords as in the example of Table 2. The entropy decoding engine 620 performs entropy decoding using this mapping information.

The above-described adaptive entropy encoding methods adaptively adjust the probability of each binary value or symbol according to the context during encoding, and thus have higher coding efficiency than the case where the probability is fixed. The reason is that when the probability is fixed, since the trained probability is applied to each symbol, a loss occurs when the probability is different from the actual occurrence probability. However, there is room for improvement in terms of coding efficiency when the adaptive entropy coding method including the aforementioned CABAC coding method is applied to a flexible video encoder. This is because when the context-based adaptive entropy coding method is applied to the flexible video coding, more accurate probability estimation can be performed using the similarity between layers.

In other words, the entropy encoding method of the present invention, when entropy encoding a symbol of a given encoding element (SE), not only the symbols of the same or other encoding elements (SE) of the same layer, but also the same or other encoding elements belonging to a lower layer ( Referencing the symbols of the SE), the more accurate statistical characteristics can be known and the entropy coding efficiency can be improved. This is because, depending on the encoding element SE, the encoding element SE to be currently encoded may be more correlated with the encoding element SE belonging to the lower layer than the encoding element SE belonging to the neighboring block of the same layer. .

To this end, the context determiner of the adaptive entropy encoder according to the present invention determines the context by referring to not only encoding elements of the same layer but also lower layer information when determining the context. Hereinafter, the operation of the context determiner of the present invention will be described in detail with reference to FIGS. 7A to 7B.

7A to 7B are reference diagrams for describing an operation of a context determiner according to the present invention.

7A to 7B, the context determiner of the present invention has the following differences compared with the context determiner of the conventional entropy encoding method. First, the context template T 710 is associated with symbols of the same or other coding element SE corresponding to a lower layer and associated with a context template {S ₀ , S ₁ , ..., S _T-1 } that is used in the related art. Additional information corresponding to the attribute includes {S _T , S _{T + 1} , ..., S _{T + K-1} }. For example, as shown in FIG. 7B, if the layer to which the current block 731 belongs is the Mth layer 730, the current together with the symbols {S ₀ , S ₁ } of neighboring blocks belonging to the same M layer 730. Symbols {S _T , S _{T + 1} , ??, S _{T + K-1} } present in each lower layer from the M-1 layer 740 to the 0th layer 750 which are lower than the layer may also be present. Include as context template 710. As symbols corresponding to lower layers are added to the context template 710, new contexts are added to the contexts {C ₀ , C ₁ , ..., C _N-1 } that are also used in the context set C 720. {C _N , C _{N + 1} , ..., C _{N + Q-1} } is added. Meanwhile, the blocks of the lower layer include blocks corresponding to the same position as the current layer and neighboring blocks around the same. It is desirable to. Referring to FIG. 7B, the symbol S _T 741 of the M−1 _th layer 740 and the symbol S _{T + K-5} 751 of the 0 _th layer 750 correspond to the same location as the current block 731. Are the symbols of the block. When the image size of the current layer and the lower layer are different, when the image is scaled by up-sampling or down-sampling, a block corresponding to the position of the current block is found. It is preferred to be determined.

The layer may be a temporal layer, a spatial layer, an SNR layer, or a mixture of three types according to the type of flexible video encoding.

8A to 8G are diagrams illustrating an embodiment of an adaptive entropy encoder using a context determiner according to the present invention.

8A illustrates a case where CABAC is used as an entropy encoding method, FIG. 8B illustrates a case where a general arithmetic encoding method is used, and FIG. 8C illustrates a case where an encoding method using mapping information between symbols and codewords is used. Is a block diagram of a CABAC encoder when lower layer information is also used for binarization.

In the entropy encoder of FIG. 8A, the binary string converter 810 converts a symbol of an encoding element SE to be encoded into a binary string. The context determiner 812 determines a context for each binary value of the converted binary string. When determining the context corresponding to each binary value, not only the encoding element information belonging to the same layer but also the encoding element information belonging to the lower layer is read from the lower layer encoding element storage unit 814 and added to the context template to determine the context. use. The probability estimator 816 finds a probability value with the context determined by the context determiner 812, and the arithmetic coding engine 818 receives a probability value and a binary value, respectively, and performs arithmetic encoding.

In the entropy encoder of FIG. 8B, unlike the entropy encoder using the CABAC encoding method of FIG. 8A, a process of converting a symbol value to a binary string is not necessary. Therefore, the binary string converter 810 is omitted in the entropy encoder of FIG. 8A. . Therefore, the context determining unit 820 of FIG. 8B reads encoding element information belonging to a lower layer from the lower layer encoding element storage unit 822 as well as encoding element information belonging to the same layer, and adds the encoding element information to a context template to encode. Determine the context for the symbol value. The probability estimator 824 finds a probability value with the context determined by the context determiner 820, and the arithmetic coding engine 826 performs arithmetic coding on the input symbol value.

In the entropy encoder of FIG. 8C, the context determiner 830 reads not only the encoding element information belonging to the same layer but also the encoding element information belonging to the lower layer from the lower layer encoding element storage unit 832 and adds it to the context template. Determine the context for the symbol value to be encoded. The determined context is used in the process of selecting one mapping relationship among a plurality of mapping relationships in the symbol-code mapping information storage unit 834. The entropy coding engine 836 has a mapping relationship between symbols and code words, and outputs corresponding code words for the input symbols.

FIG. 8D illustrates an entropy encoder when encoding element information belonging to a lower layer is used in the process of binarizing a symbol in the binary string converter 840 as well as the context determiner 842. That is, when encoding a symbol for a given encoding element, the binary string conversion unit 840 of the CABAC encoder uses a symbol of the same or another encoding element of the same or lower layer or information indicating the attribute of the encoding element. Converts to a binary string. In other words, in converting to a binary string, only encoding elements of the same layer may be referred to, encoding elements of the same layer and encoding elements of a lower layer may be referred together, or only encoding elements of a lower layer may be referred to. . In addition, the lower layer encoding element storage 844, the context determination unit 842, the probability estimation unit 846, and the arithmetic coding engine 848 are the same as those of FIG. 8A.

On the other hand, as described above, in the method of adaptively converting the relation itself converted to a binary string by referring to encoding elements of the same layer, the same layer and the lower layer, or only the lower layer, the context determination unit encodes the lower layer or the same layer. Regardless of whether or not an element is referenced, it has an effect of increasing encoding efficiency by itself. Preferred embodiments for this are disclosed in FIG. 8E. That is, the context determiner 852 refers to only encoding element information of the same layer as described in FIG. 2, and the binary string converter 850 encodes only encoding elements of the same layer, or the same layer and a lower layer, or a lower layer. See. Accordingly, the hierarchical encoding element storage unit 854 stores encoding elements of only the same layer, only the same layer and the lower layer, or only the lower layer according to the above-mentioned reference information selection.

As a specific implementation example of the binary string converters 840 and 850, a plurality of mapping tables may be stored between a symbol value of a symbol and a binary string to select one mapping table from each table according to the selection of the reference information, or a symbol The mapping relationship between the symbol value and the binary value may be a fixed table but a method of remapping the relationship between the symbol and the symbol value according to the selection of the reference information. The former is to vary the binary string value itself for the symbol. For example, a new table is constructed with a binary value of 100 for the mb_type symbol value 1 as a different value. The latter method differs only when the binary string for symbol value 1 is 100 and the binary string for symbol value 2 is 101. Set the binary string for symbol value 1 to 101 and the binary string for symbol value 2 to 100.

8A to 8B, the context determiner 812 and 820 and the probability estimator 818 and 824 may be implemented separately, but another embodiment may be implemented. 8F to 8G, the function of the probability estimator 818 is combined with the context determiner 812 of FIG. 8A to implement a new content determiner 860, and the context determiner of FIG. 8B. The function of the probability estimator 824 may be added to 820 to implement a new context determiner 870. 8A to 8B, the probability estimators 818 and 824 determine specific probability values based on the context determined by the context determiner 812 and 820. However, in FIGS. 8F to 8G, the context determiner 860 or FIG. 870 itself determines a certain probability value. In this case, the context determination units 860 and 870 also perform a function of probability update.

Of course, the same modifications are also possible in FIGS. 8D and 8E in actual implementation.

9A to 9G are diagrams illustrating an embodiment of an adaptive entropy decoder using a context determiner according to the present invention.

In more detail, FIGS. 9A to 9D are configuration diagrams of an entropy decoder corresponding to the entropy encoder of FIGS. 8A to 8D, respectively. Accordingly, FIG. 9A illustrates a case where CABAC is used as an entropy decoding method, FIG. 9B illustrates a case where a general arithmetic decoding method is used, and FIG. 9C illustrates a case where a decoding method using mapping information between symbols and codewords is used. 9D is a configuration diagram of a CABAC decoder when lower layer information is also used for binarization.

In the entropy decoder of FIG. 9A, the context determiner 910 determines a context for a binary value to be currently decoded from an entropy coded bit string, according to each position of a binary value in a binary string to be decoded. Determine the context. When determining the context corresponding to each binary value, not only the encoding element information belonging to the same layer but also the encoding element information belonging to the lower layer is read from the lower layer encoding element storage unit 911 and added to the context template to determine the context. use. The determined context is used by the entropy decoding unit 912 to decode a binary value from the input bit string. The probability estimator 913 receives a context and estimates a probability of 0 and 1 for each binary value. The arithmetic decoding engine 914 sequentially decodes a binary value from the bit string with the estimated probability value and transfers the result to the probability estimator 913 to update the probability of the binary value. Binary values sequentially decoded by the arithmetic decoding engine 914 are converted into symbol data through the symbol value converter 915.

On the other hand, unlike the CABAC encoder described above, a general arithmetic decoder in which the arithmetic decoding engine directly generates a symbol value has the same structure and function as the CABAC decoder except for the symbol value converter 915. However, specific embodiments may vary, so reference numerals are indicated differently.

In the entropy decoder of FIG. 9C, the context determiner 920 has a symbol value instead of a binary value of a symbol, and encodes belonging to a lower layer from the lower layer encoding element storage 921 as well as encoding element information belonging to the same layer. Read the element information and add it to the context template to determine the context. The determined context is used in the process of selecting one mapping relationship among a plurality of mapping relationships in the symbol-code mapping information storage unit 923. The entropy decoding engine 924 has a mapping relationship between the selected symbol and the codeword, and outputs a symbol corresponding to the codeword.

In the entropy decoder of FIG. 9D, the encoding element information belonging to the lower layer encoding element stored in the lower layer encoding element storage 931 stores the binary value in the symbol value converter 932 as well as the context determiner 930. It is used in the process of converting to a symbol value. The detailed process is the same as the reverse process of the binary string conversion process described with reference to FIG. 8D.

On the other hand, as described above, in the method of adaptively converting a relationship itself to a symbol value by referring to encoding elements of the same layer, the same layer and a lower layer, or only a lower layer, the context decision unit encodes the lower layer or the same layer. Regardless of whether or not an element is referenced, it has an effect of increasing encoding efficiency by itself. A preferred embodiment of this is disclosed in FIG. 9E. That is, the context determiner 940 refers only to encoding element information of the same layer as described with reference to FIG. 4, and the symbol value converter 942 encodes only encoding elements of the same layer, or the same layer and a lower layer, or a lower layer. See. Accordingly, the hierarchical encoding element storage unit 944 stores encoding elements of the same layer, the same layer and the lower layer, or the lower layer according to the above-mentioned reference information selection.

9A to 9B, the context determiner 910 and 916 and the probability estimator 913 and 918 may be implemented separately, but in another embodiment, FIGS. 9F to 9B. As illustrated in FIG. 9G, the function of the probability estimator 913 may be combined with the context determiner 910 of FIG. 9A to implement a new context determiner 950. Similarly, the function of the probability estimator 918 may be combined with the context determiner 916 of FIG. 9B to implement the new context determiner 960. 9A to 9B, the probability estimators 913 and 918 determine specific probability values based on the context determined by the context determiner 910 and 916. However, in FIGS. 9F to 9G, the context determiner 950 and 960 itself determine specific probability values. Determine. In this case, the context determination units 950 and 960 also perform a function of probability update.

10A to 10C are flowcharts of an entropy encoding method according to an embodiment of the present invention.

A symbol of an encoding element to be encoded is received (S1010). In some cases, the symbol value of the input encoding element is converted into a binary string. An example of conversion to a binary string is as described above with reference to Table 1. Then, a context corresponding to each binary value of the symbol value or binary string of the encoding element is determined (S1012). At this time, the context is determined by referring to encoding elements of the same layer and lower layer as the layer of the block to which the encoding element to be encoded belongs, or by referring to only the encoding elements of the lower layer. Determination of context is as described above with reference to FIGS. 7A-7B.

According to the selected context, a probability for the coding element is estimated (S1014), and entropy encoding is performed using the estimated probability value. If it is converted to a binary string for entropy encoding as described above with reference to Table 1, the probability of the value of 0 and the value of 1 is estimated, and using the estimated probability value, the symbol value of the input encoding element or The binary value is entropy encoded (S1016). In addition, the probability model is updated according to the symbol value or the binarized value of the entropy-encoded encoding element (S1018). In some cases, two stages of determining the context (S1012) and probability estimating (S1014) are performed in one step. The same layer as the layer of the block to which the encoding element to be encoded belongs is determined without explicit context determination. And a probability of the encoding element may be estimated by referring to the encoding element of the lower layer or by referring to only the encoding element of the lower layer.

Referring to FIG. 10B, an entropy encoding method according to another embodiment of the present invention is described. After receiving a symbol of an encoding element to be encoded (S1020), a context corresponding to the encoding element is determined (S1022). At this time, the description of the step of determining the context (S1022) is the same as the context determination step (S1012) described above with reference to Figure 10a. According to the determined context, as described above with reference to FIG. 8C, one mapping relationship is selected from the plurality of symbol-code mapping relationships (S1024). Thereafter, entropy encoding is performed on the input encoding element according to the selected mapping relationship (S1026), and a corresponding codeword is output. In addition, even in the case of FIG. 10B, two steps of determining a context (S1022) and selecting a mapping relationship (S1024) are performed in one step, and the same as the layer of the block to which an encoding element to be encoded belongs without determining an explicit context. The encoding element of the layer and the lower layer may be referred to, or only the encoding element of the lower layer may be referred to to select a symbol-code mapping relationship for encoding the encoding element.

Referring to FIG. 10C, when a CABAC entropy encoding method is used when information of a same layer or a lower layer is also used for binarization, a symbol of an encoding element to be encoded is received (S1030). The symbol value of the input encoding element is converted into a binary string (S1032). In the binary string conversion step, a binary string is determined according to encoding element information corresponding to the same layer or a lower layer. That is, when binarizing a symbol for a given encoding element, information indicating a symbol or attribute of an encoding element or a symbol of the same or another encoding element in the same layer, the same layer and the lower layer, or the lower layer is used. Converts the symbol to a binary string. In other words, in converting to a binary string, only encoding elements of the same layer may be referred to, encoding elements of the same layer and encoding elements of a lower layer may be referred together, or only encoding elements of a lower layer may be referred to. .

Then, a context corresponding to each binary value of the symbol of the binarized encoding element is determined (S1034). In this case, the context is determined by referring to encoding elements of only the same layer as the layer of the block to which the encoding element to be encoded, referring to encoding elements of the same layer and lower layer, or referring only to encoding elements of the lower layer. Determination of context is as described above with reference to FIGS. 7A-7B.

According to the selected context, a probability for the coding element is estimated (S1036), and entropy encoding is performed using the estimated probability value. If it is converted to a binary string for entropy encoding as described above with reference to Table 1, the probability of the value of 0 and the value of 1 is estimated, and the encoding element is entropy encoded using the estimated probability value (S1038). ). In addition, the probability model is updated according to the symbol value or the binarized value of the entropy-encoded encoding element (S1040). In some cases, two steps of determining the context (S1034) and probability estimating (S1036) are performed in one step. The same layer as the layer of the block to which an encoding element to be encoded belongs is included without selecting an explicit context. And a probability of the encoding element may be estimated by referring to the encoding element of the lower layer or by referring to only the encoding element of the lower layer.

11A to 11C are flowcharts of an entropy decoding method according to an embodiment of the present invention.

Referring to FIG. 11A, when describing an entropy decoding method using CABAC, a compressed bitstream is input (S1110), and a context is determined according to each position of a binary value in a binary string to be decoded (S1112). . When determining the context corresponding to each binary value (S1112), as described above with reference to FIG. 9A, not only encoding element information belonging to the same layer but also encoding element information belonging to a lower layer are referred to. Determination of such a context is as described above with reference to FIGS. 7A to 7B. According to the determined context, a probability for each binary value 0 and 1 is estimated (S1114). Entropy decoding for sequentially decoding binary values from the compressed bit string using the estimated probability is performed (S1116). In addition, by using the decoded binary values, the probability model for each binary value of 0 and 1 is updated (S1118). Binary values decoded through entropy decoding are converted into symbol values (S1119). The symbol value conversion step S1119 is as described above with reference to the symbol value conversion unit 915 of FIG. 9A.

Meanwhile, unlike the entropy decoding method using the CABAC, the method of FIG. 11A may be applied to a general arithmetic decoding method for generating a symbol value directly in the entropy decoding process (S1116). In this case, the symbol value conversion step S1119 is excluded from FIG. 11A, and the description of FIG. 11A is applied as it is. However, since the entropy decoding step S1116 does not decode the binarized value but decodes the symbol value, the context determination step S1112 determines the context for the symbol to be decoded. In addition, the probability of each symbol value is estimated according to the determined context (S1114). When determining the context (S1112), not only the encoding element information belonging to the same layer but also the encoding element information belonging to the lower layer are referred to. After performing entropy decoding (S1116) for decoding the symbol value from the compressed bit string using the estimated probability, the probability model for the corresponding symbol value is updated according to the decoded symbol value (S1118).

Referring to FIG. 11B, a decoding method using mapping information between a symbol and a codeword will be described. First, a compressed bit string is received (S1120). Then, not only the encoding element information belonging to the same layer as the symbol to be decoded but also the encoding element information belonging to the lower layer are read and added to the context template to determine the context of the symbol to be decoded (S1122). Using the determined context, one mapping relationship is selected from among the plurality of mapping relationships as described above with reference to FIG. 9C (S1124). The corresponding symbol is output with respect to the codeword of the compressed bit string having the mapping relationship between the selected symbol and the codeword (S1126).

Referring to FIG. 11C, when a CABAC decoding method is used when information of the same layer or a lower layer is also used for binarization, a compressed bitstream is input (S1130) and a binary value in a binary string to be decoded. The context is determined according to each position (S1132). According to the determined context, a probability of data of each of binary values 0 and 1 is estimated (S1134). Entropy decoding for sequentially decoding binary values from the compressed bit string using the estimated probability is performed (S1136). In addition, by using the decoded binary values, the probabilistic model for each binary value of 0 and 1 is updated (S1138). Binary values decoded through entropy decoding are converted into symbol values (S1140). In this case, the process of converting the binary values into symbol values is as described above with reference to FIG. 9E. That is, when the binary values are converted into symbol values, reference information indicating a property of a symbol or an encoding element of the same or another encoding element of the same layer, the same layer and a lower layer, or the lower layer is obtained by decoding. .

As described above, a method of adaptively converting a relationship itself to a symbol value by referring to encoding elements of the same layer, the same layer and the lower layer, or only the lower layer may be encoded in the lower layer or the same layer in the context determination step. Regardless of whether or not an element is referenced, it has an effect of increasing encoding efficiency by itself.

12 is a block diagram illustrating an example of a video encoding apparatus having an entropy encoder of the present invention.

The video encoding apparatus includes a motion estimation unit 1202, a motion compensator 1204, an intra prediction unit 1206, a transform unit 1208, a quantization unit 1210, a reordering unit 1212, and an entropy encoder 1214. ), An inverse quantizer 1216, an inverse transform unit 1218, a filter 1220, and a frame memory 1222.

The encoding apparatus encodes the macroblock of the current picture in one encoding mode selected from various encoding modes. To this end, encoding is performed under all modes that inter-prediction and intra-prediction may have, and the rate-distortion cost (RDcost) is calculated to determine the mode with the smallest value as the optimal mode. Perform the encoding. Finding the predicted value of the macro block of the current picture in the reference picture for interprediction is performed by the motion estimation unit 1202. When the reference block is found in units of 1/2 or 1/4 pixels, the motion compensator 1204 determines the reference block data value by calculating these intermediate pixel values. The inter prediction is performed by the motion estimator 1202 and the motion compensator 1204. In addition, the intra prediction that finds the prediction value of the macro block of the current picture within the current picture is performed by the intra prediction unit 1206. Is performed. Whether to perform inter-prediction or intra-prediction on the current macro block is calculated by calculating the rate-distortion cost under all encoding modes and determining the mode in which the value is the smallest as the encoding mode of the block. Perform encoding on.

If inter prediction or intra prediction is performed to find the prediction data to be referred to by the macro block of the current frame, the quantization unit 1210 performs a transformation by subtracting it from the macro block of the current picture. Perform quantization. The subtraction of the motion estimation reference block from the macroblock of the current frame is called residual. The residual value is encoded to reduce the amount of data during encoding. The quantized residual value passes through the reordering unit 112 for encoding by the entropy encoder 1214. The entropy encoder 1214 is configured as described above with reference to FIGS. 8A to 8G.

Meanwhile, in order to obtain a reference picture to be used for interprediction, the quantized picture is reconstructed through the inverse quantizer 1216 and the inverse transform unit 1218. The reconstructed current picture is stored in frame memory and used to perform interprediction on the next picture. When the reconstructed picture passes the filter 1220, it becomes a picture including some encoding error in the original picture. [0077] Fig. 13 is a block diagram showing an example of a video decoding apparatus having an entropy decoder of the present invention.

The video decoding apparatus includes an entropy decoder 1310, a reordering unit 1320, an inverse quantization unit 1330, an inverse transform unit 1340, an intra prediction unit 1350, a motion compensator 1360, and a filter 1370. And a frame memory 1380.

When the bit stream encoded by the video encoding apparatus is input, the entropy decoder 1310 extracts symbol data by entropy decoding. The following components have the same functions as the components described above with reference to FIG. 12.

The entropy encoding and decoding method described above can be created by a computer program. Codes and code segments constituting the program can be easily inferred by a computer programmer in the art. In addition, the program is stored in a computer readable media, and read and executed by a computer to implement an entropy encoding and decoding method. The information storage medium includes a magnetic recording medium, an optical recording medium, and a carrier wave medium.

So far I looked at the center of the preferred embodiment for the present invention. Those skilled in the art will appreciate that the present invention can be implemented in a modified form without departing from the essential features of the present invention. Therefore, the disclosed embodiments should be considered in descriptive sense only and not for purposes of limitation. The scope of the present invention is shown in the claims rather than the foregoing description, and all differences within the scope will be construed as being included in the present invention.

As described above, according to the present invention, encoding efficiency can be improved by performing entropy encoding using not only the encoding elements of the same layer but also the encoding elements of the lower layer in the flexible video encoding.

Claims (45)

(a) determining a context by referring to encoding elements of the same layer and lower layer as the layer of the block to which the encoding element to be encoded belongs, or referring only to encoding elements of the lower layer; And and (b) entropy-encoding the encoding element using the determined context. The method of claim 1, The encoding element referred to in the step (a) is a hierarchical entropy encoding method, characterized in that the encoding element of the same kind as the encoding element to be encoded. The method of claim 1, And the context indicates a probability value. The method of claim 1, And the context indicates one of a plurality of predetermined symbol-codeword relationships. The method of claim 1, The coding element referred to in the step (a) includes a coding element of the same kind as the coding element to be encoded, as well as other types of coding elements. The method of claim 1, wherein the entropy encoding in step (b) A hierarchical entropy encoding method using arithmetic encoding using a predetermined arithmetic expression, or encoding using a relationship between a symbol and a codeword. The method of claim 1, The entropy encoding in step (b) has a plurality of predetermined mapping information between symbol-codewords and is performed using the entropy encoding method. The method of claim 6, wherein the entropy encoding in step (b) A hierarchical entropy encoding method comprising binary arithmetic coding, arithmetic coding, Huffman coding, or single variable length coding. The method of claim 1, (a0) hierarchical entropy encoding method further comprising converting the encoding element to be encoded into a binary string. The method of claim 9, wherein step (a0) The encoding element to be encoded is referred to by referring to an encoding element of the same layer as the layer of the block to which the encoding element is to be encoded, referring to encoding elements of a lower layer, or referring to encoding elements of the same layer and a lower layer. And converting the binary string into a hierarchical entropy encoding method. The method of claim 10, wherein in step (a0) And a mapping relationship between an encoding element to be encoded and a binary string is fixed or variably according to a context determined by the encoding element to be referred to. (a) The encoding to be encoded by referring to encoding elements of the same layer as the layer of the block to which the encoding element to be encoded belongs, referring to encoding elements of a lower layer, or referring to encoding elements of the same layer and lower layer. Converting the element to a binary string; (b) determining a context by referring to encoding elements of the same layer as the layer of the block to which the encoding element to be encoded belongs; And (c) entropy encoding the encoding element by using the determined context. (a) receiving the entropy-coded data and determining a context by referring to encoding elements of the same layer and lower layer as the layer of the block to which the entropy-encoded encoding element belongs, or referring only to encoding elements of the lower layer; And and (b) entropy decoding the entropy-encoded data using the determined context. The method of claim 13, The encoding element referred to in the step (a) is a hierarchical entropy decoding method, characterized in that the encoding element of the same kind as the encoded encoding element. The method of claim 13, And the context indicates a probability value. The method of claim 13, And the context indicates one of a plurality of predetermined symbol-codeword relationships. The method of claim 13, wherein the encoding element referred to in the step (a) And a coding element of the same kind as the encoded coding element, as well as other types of coding elements. The method of claim 13, wherein the entropy decoding in step (b) An arithmetic decoding method using a predetermined arithmetic expression, or a decoding method using a symbol-symbol relationship. 19. The method of claim 18, wherein entropy decoding in step (b) A hierarchical entropy decoding method comprising binary arithmetic decoding, arithmetic decoding, Huffman decoding, or single variable length decoding. The method of claim 13, and (c) converting the decoded binary string into a symbol value. The method of claim 20, wherein step (c) The entropy decoded binary string is referred to as a symbol by referring to an encoding element of the same layer as a layer of a block to which an encoded encoding element belongs, referring to encoding elements of a lower layer, or referring to encoding elements of the same layer and a lower layer. Hierarchical entropy decoding method characterized in that the step of transforming. (a) receiving entropy-coded data and determining a context by referring to encoding elements of the same layer as a layer of a block to which the entropy-encoded encoding element belongs; (b) entropy decoding the entropy encoded data using the determined context; And (c) the entropy decoded binary by referring to an encoding element of the same layer as the layer of the block to which the encoded encoding element belongs, referring to an encoding element of a lower layer, or referring to all of encoding elements of the same layer and lower layer. Converting the string into a symbol. The method of claim 22, wherein in step (c) And a mapping relationship between a binary string and a symbol is fixed or variably according to a context determined according to the encoding element referenced. A context determination unit for determining a context by referring to encoding elements of the same layer and lower layer as the layer of the block to which the encoding element to be encoded belongs, or referring only to encoding elements of the lower layer; And And an entropy encoding engine that entropy encodes the encoding elements by using the determined context. The method of claim 24, And the context determiner determines a context by referring to an encoding element of the same type as the encoding element to be encoded. The method of claim 24, And the context indicates a probability value. The method of claim 24, And the context indicates one of a plurality of predetermined symbol-codeword relationships. The method of claim 24, wherein the context determination unit The hierarchical entropy encoding apparatus of claim 1, wherein the context is determined by referring to not only the encoding element of the same kind as the encoding element to be encoded but also other encoding elements. 25. The method of claim 24, wherein the entropy coding engine An arithmetic encoding using a predetermined arithmetic expression, or a hierarchical entropy encoding apparatus characterized by performing the encoding using the relationship between the symbol-code. 30. The method of claim 29, wherein the entropy coding engine A hierarchical entropy encoding apparatus characterized by performing binary arithmetic coding, arithmetic coding, Huffman coding, or single variable length coding. The method of claim 24, And a binary string converter configured to convert an encoding element to be encoded into a binary string. 32. The binary string converter of claim 31, wherein Binary encoding element to be encoded by referring to an encoding element of the same layer as the layer of the block to which the encoding element to be encoded belongs, referring to encoding elements of a lower layer, or referring to encoding elements of the same layer and lower layer. Hierarchical entropy encoding apparatus, characterized in that the conversion to a string. Binary encoding element to be encoded by referring to an encoding element of the same layer as the layer of the block to which the encoding element to be encoded belongs, referring to encoding elements of a lower layer, or referring to encoding elements of the same layer and lower layer. A binary string converting unit converting the string into strings; A context determination unit that determines a context by referring to encoding elements of the same layer as the layer of the block to which the encoding element to be encoded belongs; And And an entropy encoding engine that entropy encodes the encoding element by using the determined context. 34. The apparatus of claim 33, wherein the binary string conversion unit An entropy encoding apparatus according to a context determined according to a coding element to be referred to, is converted into a binary string using a relation in which a mapping relationship between a coding element to be encoded and a binary string is fixed or variably changed. A context determination unit which receives entropy-coded data and determines a context by referring to encoding elements of the same layer and lower layer as the layer of the block to which the entropy-encoded encoding element belongs, or referring only to encoding elements of the lower layer; And And an entropy decoding engine that entropy decodes the entropy-encoded data using the determined context. 36. The method of claim 35 wherein And the context determiner determines a context by referring to an encoding element of the same type as the encoded encoding element. 36. The method of claim 35 wherein And the context indicates a probability value. 36. The method of claim 35 wherein And the context indicates one of a plurality of predetermined symbol-codeword relationships. 36. The apparatus of claim 35, wherein the context determiner And a context element determined by referring to not only the encoding element of the same type as the encoded encoding element but also other kinds of encoding elements. 36. The method of claim 35, wherein the entropy decoding engine And performing arithmetic decoding using a predetermined arithmetic expression or decoding using a relation between symbol and codewords. 41. The method of claim 40, wherein the entropy decoding engine A hierarchical entropy decoding apparatus comprising binary arithmetic decoding, arithmetic decoding, Huffman decoding, or single variable length decoding. 36. The method of claim 35 wherein And a symbol value converting unit converting the decoded binary string into a symbol value. 43. The apparatus of claim 42, wherein the symbol value converting unit Symbolize the entropy decoded binary string by referring to an encoding element of the same layer as the layer of the block to which the encoded encoding element belongs, referring to an encoding element of a lower layer, or referring to all encoding elements of the same layer and a lower layer. Hierarchical entropy decoding apparatus characterized in that the conversion. A context determination unit which receives entropy-coded data and determines a context by referring to encoding elements of the same layer as a layer of a block to which the entropy-encoded encoding element belongs; An entropy decoding engine that entropy decodes the entropy coded data using the determined context; And Symbolize the entropy decoded binary string by referring to an encoding element of the same layer as the layer of the block to which the encoded encoding element belongs, referring to an encoding element of a lower layer, or referring to all encoding elements of the same layer and a lower layer. Entropy decoding apparatus comprising a symbol value conversion unit for converting. 45. The apparatus of claim 44, wherein the symbol value converting unit Entropy decoding apparatus for converting a binary string into a symbol value by using a relation in which a mapping relationship between an encoded encoding element and a binary string is fixed or variably according to a context determined by the encoding element to be referred to .

Images