KR101136858B1 - Encoder and encoding method in video coding standard - Google Patents

Abstract

The present invention relates to an encoding apparatus and method in the H.264 / AVC video compression standard. An encoding method according to an embodiment of the present invention creates a memory structure capable of effectively storing a VLCT used in Coeff_token encoding. That is, codewords existing in the VLCT are divided into length information and value information, and the length information and the value information of the codewords are stored in each allocated memory area. The size of the length information of the codeword is configured to be reduced, and the length information of the reduced codeword is stored in the allocated memory area. The encoding method according to the embodiment of the present invention can improve the performance of the encoding system and reduce the implementation cost of the encoding system by using a reduced size encoding table.

Description

ENCODER AND ENCODING METHOD IN VIDEO CODING STANDARD}

An embodiment of the present invention relates to a data compression apparatus, and more particularly, to an encoding apparatus and method in the H.264 / AVC video compression standard.

Recently, multimedia information, in which image and sound are combined with communication and computers, is integrated into new media. For example, as a high-speed data transmission network is provided, users can watch stereo sound and high-definition video, and can face to face with a distant person through a videophone. In addition, it is possible to purchase goods while viewing product information in real time through a computer or a TV, and to enjoy music or a movie through a website. It is also possible to take video lectures through a computer.

Such multimedia information has been developed based on video compression technology. Data that conveys information can be compressed by removing redundant elements from the data (elements that are not necessary to accurately restore the data). In the case of lossless compression, statistical redundancy is removed to restore the data recovered at the decoder to exactly match the original data. However, the current lossless compression method has a maximum compression efficiency of about 3 to 4 times, which is not high in compression efficiency. Most compression techniques are based on lossy compression. In the case of lossy compression, the data reconstructed at the decoder is not the same as the original data, but subjective redundancy is removed to obtain high compression efficiency. In image or video compression, subjective redundancy is a removable element that does not significantly affect the image quality that the viewer can intuitively feel.

The H.264 / AVC video compression standard, one of various video compression standards, has superior compression performance compared to the existing video compression standards. The H.264 / AVC video compression standard supports bidirectional streaming or bidirectional broadcast. Thus, the H.264 / AVC video compression standard provides video communications, cable, satellite, terrestrial broadcast, or bidirectional storage for data storage devices, DVDs, and other storage media, Ethernet, modems, and local area networks. It is applied to a wide range of applications such as conversation service and multimedia streaming service through mobile network or mobile network.

Embodiments of the present invention provide a method for storing a reduced sized encoding table in the H.264 / AVC video compression standard.

In the encoding method of the context-adaptive variable length code method according to an embodiment of the present invention, the encoding method includes information on the number of non-zero coefficients (Tc) of the quantized transform coefficients and the number of coefficients (T1s) having a size of one. Generating a parameter value codenum with reference; Calculating a default value of a bit stream length of a codeword with reference to the parameter value (codenum); Reading an offset value from a storage unit according to an address (Add) and an index (Idx) value generated by referring to the parameter value codenum; Generating bit stream length information of a codeword according to the base and the offset; And generating a bit stream of the codeword using the bit stream value information of the codeword and the bit stream length information of the generated codeword.

An encoding apparatus according to another embodiment of the present invention includes a codenum generator for generating a parameter from quantized transform coefficients; A default value calculator for calculating a default value according to the generated parameter value; A storage unit for storing a bit stream length information value of a variable length codeword and a difference value of the default value; And an adder configured to generate bit stream length information of the variable length codeword by adding the calculated default value and a difference value read according to the generated parameter value.

The encoding apparatus and method according to an embodiment of the present invention can improve the performance of the encoding system and reduce the implementation cost of the encoding system by using the reduced size encoding table.

1 is a block diagram illustrating an encoder of an H.264 / AVC compression standard according to an embodiment of the present invention.
2 is a diagram illustrating a process of encoding quantized coefficients.
3A to 3C are diagrams schematically illustrating a variable length code table referred to in a context adaptive variable length code encoding scheme.
4 is a diagram illustrating a memory structure of a variable length code table according to a first embodiment of the present invention.
5 is a block diagram illustrating an entropy encoder according to an embodiment of the present invention.
FIG. 6 is a diagram illustrating a correlation between bit stream length (LEN) information of a codeword and a parameter codenum according to an embodiment of the present invention.
7 is a diagram illustrating a process of setting a default value for a parameter (codenum) according to an embodiment of the present invention.
8 is a diagram illustrating a memory structure for a variable length code table according to a second embodiment of the present invention.
9 is a flowchart illustrating an operation of encoding Coeff_token in a memory structure according to an embodiment of the present invention.

Advantages and features of the present invention, and methods for achieving the same will be described with reference to embodiments described below in detail with the accompanying drawings. However, the present invention is not limited to the embodiments described herein and may be embodied in other forms. The embodiments are provided so that those skilled in the art can easily carry out the technical idea of the present invention to those skilled in the art.

In the drawings, embodiments of the present invention are not limited to the specific forms shown and are exaggerated for clarity.

Although specific terms are used herein. It is used for the purpose of illustrating the present invention and is not intended to limit the scope of the present invention as defined in the meaning limitations or claims.

The expression " and / or " is used herein to mean including at least one of the elements listed before and after. In addition, the expression “connected / combined” is used to include directly connected to or indirectly connected to other components.

In this specification, the singular forms also include the plural unless specifically stated otherwise in the phrases. Also, as used herein, "comprising" or "comprising" means to refer to the presence or addition of one or more other components, steps, operations and elements.

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

1 is a block diagram illustrating an encoder of an H.264 / AVC compression standard according to an embodiment of the present invention.

The H.264 / Advanced video coding (AVC) standard is a standard for the encoded representation of image information. The H.264 / AVC standard basically defines two things. First, an encoded representation (or syntax) representing the compressed image data is defined, and second, a method of decoding the syntax to recover the image information is defined. That is, the H.264 / AVC standard does not define codecs (enCOder / DECoder, CODEC) separately, but defines the syntax of an encoded video bit stream and a method of decoding such a bit stream. In general, an encoder compatible with the H.264 / AVC standard will include the blocks shown in FIG.

Referring to FIG. 1, the encoder of the H.264 / AVC video compression standard includes a motion estimation block 110, a motion compensation block 120, an intra prediction block 130, And a filter 140, a reorder 150, and an entropy encoder 160.

The motion compensation method used in video compression is to compensate for the motion of a rectangular section or block of the current frame. The region of the reference frame is searched to find an M × N sample block that matches the M × N (M, N is a natural number) sample block of the current frame. Here, the reference frame is a past or future frame or a frame previously encoded and transmitted. The process of searching an area of a reference frame compares all or some of the M × N blocks of the current frame with the possible M × N blocks in the search area (usually the area relative to the current block's location), and matches It is the process of finding a domain. In other words, the candidate region having the minimum error generated by the difference between the current M × N block and the candidate region is selected as the most matching region. This operation is performed by the motion estimation unit 110. The selected candidate region becomes a prediction block for the current M × N block, and an M × N error block is created by the difference between the current block and the prediction block. This operation is performed by the motion compensator 120.

The encoder 100 processes the input frame F _n in macroblock units. Each macroblock is encoded in intra mode or inter mode. Intra mode is a method of encoding without motion compensation. Inter mode is a method of encoding using motion compensated prediction. The predictive block P for each block in the macroblock is generated by the reconstructed block. In intra mode, the prediction block P is generated by comparing the current block with a previously encoded block (or a decoded and reconstructed block). The generation operation of the prediction block P in the intra mode is performed by the intra predictor 130. In the inter mode, the prediction block P is generated by predicting motion compensation from the reference block. In FIG. 1 the reference block is shown as a previously encoded frame F ′ _{n −1} .

An error block D _n is generated by the difference between the current block and the prediction block P. The generated error block D _n is transformed to another domain (T) and represented as a transform coefficient. The transform coefficients are quantized (Q) to become quantized transform coefficients (X). The quantized transform coefficients X are zigzag scanned and rearranged through the rearrangement unit 150. The relocated transform coefficients are entropy encoded by entropy encoder 160. The entropy encoded coefficients are compressed with additional data (eg, prediction mode, quantization parameters, motion vector information, etc.) needed to decode each block in the macroblock. Compressed data forms a bit stream and is transmitted or stored through a network abstraction layer (NAL).

The encoder 100 not only encodes and transmits each block in the macroblock, but also decodes the encoded data again to generate reference data for later prediction. The quantized transform coefficient X is inversely quantized (Q ⁻¹ ) and inverse transformed (T ⁻¹ ) to produce an error block D ′ _n . The prediction block P is added to the error block D ' _{n to} produce a decoded frame uF' _n . When the decoded frame uF ' _n is filtered through the filter 140, the distortion is reduced to produce a decoded reference frame F' _n .

The H.264 / AVC standard defines profiles that support specific functions. One of the baseline profiles is mobile video terminal services such as portable multimedia player (PMP) and personal digital assistant (PDA) and video conferencing. The baseline profile supports intra coding, inter coding and entropy coding using context-based adaptive variable length code (CAVLC).

In the CAVLC encoding scheme, quantized transform coefficients generated in a 4 × 4 block are represented by five syntax elements Coeff_token, Sign of T1s, Level, Total_zeros, and Run_before, and then each syntax element is encoded. In this case, Coeff_token, Total_zeros, and Run_before syntax elements are encoded with reference to a defined variable length code table (hereinafter, referred to as 'VLCT'). The entropy encoder 160 according to an embodiment of the present invention divides a bit stream of a codeword existing in the VLCT into length information and value information. The length information and the value information of the bit stream are stored in each allocated memory area. The size of the bit stream length information is divided into a default value and an increment value, and only the increment value is stored in the memory device. The default value is calculated according to the default value calculator, and since only the increment value is stored in the memory device, the memory area in which the VLCT is stored is reduced. In addition, since the memory device area to be read is reduced, the operation speed of the encoder 100 is improved. The operation of the entropy encoder 160 will be described in detail with reference to FIGS. 5 and 9 described below.

2 is a diagram illustrating a process of encoding quantized coefficients.

In the baseline profile of the H.264 / AVC standard, quantized transform coefficients of 4x4 blocks are zigzag scanned and rearranged. The rearranged transform coefficients are encoded by the five syntax elements described above. The five syntax elements are as follows: Coeff_token is an element in which the number of coefficients other than '0' (hereinafter referred to as 'Tc') and the number of coefficients having a size of '1' in the high frequency component (hereinafter, 'T1s') are encoded. Level is an element whose size is encoded as a non-zero coefficient except for T1s. Total_zeros is an element in which the number of coefficients of size '0' existing between the first coefficient and the last non-zero coefficient in the zigzag order is encoded. Run_before is an element in which the number of coefficients of '0' existing in front of each non-zero coefficient in the reverse order of the zigzag scan is encoded.

The process of encoding the quantized transform coefficients of the 4x4 block of FIG. 2 is as follows. Rearranging the quantized coefficients of a 4x4 block by zigzag scan is 0, -3, 0, 1, -1, -1, 0, 1, 0, ... The rearranged transform coefficients are expressed as five syntax elements defined in CAVLC encoding. Tc, the number of coefficients other than '0', is '5' (-3, 1, -1, -1, 1). T1s, the number of coefficients whose magnitude is '1', that is, the absolute value is '1', is '3' (-1, -1, 1). Here, only T1s are encoded up to three from the last coefficient. Coeff_token consisting of Tc and T1s is encoded with reference to the VLCT. Referring to the VLCT shown briefly in FIGS. 3A to 3C, Coeff_token is '0000100' (Tc = 5, T1s = 3).

Sign of T1s in which the sign of T1s is encoded in the reverse order of zigzag scan is '011' (+,-,-). Here, the positive sign (+) is encoded as a '0' value, and the negative sign (-) is encoded as a '1' value. The level at which the magnitude of the coefficient other than '0' except T1s is encoded is '10010' (-3, 1). The total_zeros in which the number of coefficients of size '0' existing between the first coefficient and the last non-zero coefficient in the zigzag order is encoded is '111' (three). Run_before, in which the number of coefficients of '0' existing before each non-zero coefficient in the zigzag scan order is encoded, is '101101'. Therefore, the encoded final bit stream is '000010001110010111101101'.

3A to 3C schematically illustrate a variable length code table (VLCT) referred to in a context adaptive variable length code encoding scheme.

The quantized transform coefficients are encoded by the five syntax elements described above. One fixed length code table and three variable length code tables VLCT are used to encode Coeff_token among syntax elements. 3A-3C are VLCTs for Coeff_token, one of the three tables is selected with reference to the Tc value of the previously restored neighboring block. VLCT0 is schematically shown in FIG. 3A, VLCT1 in FIG. 3B and VLCT2 in FIG. 3C.

3A to 3C, codewords correspond to pairs of Tc and T1s. Each codeword consists of a string of bits ranging from 1 bit to 16 bits. Each bit string represents a value from 1 to 15. Codewords are stored in a memory device. Accordingly, a memory area (hereinafter referred to as 'LEN_M') and a bit stream value (hereinafter referred to as 'VAL') for storing the bit stream length (hereinafter referred to as 'LEN') of the codewords of FIGS. 3A to 3C. ), A memory area (hereinafter referred to as 'VAL_M') is required.

4 is a diagram illustrating a memory structure of a variable length code table according to a first embodiment of the present invention.

Referring to FIG. 4, codewords of VLCT0 illustrated in FIG. 3A are classified into LEN and VAL, and correspond to a pair of an address (add) and an index (idx). Memory addresses (add) for referring to LEN and VAL stored in the memory area are represented by Equations 1 to 4 using Tc and T1s. In addition, the index (idx) for referencing the LEN and VAL is expressed as in Equation (5). In Equation 1, the "<<" operator is an operator used in programming code (for example, C language) and means a shift operation. In addition, in Equation 5, the "%" operator is an operator used in programming code (for example, C language), and means the remaining operation. These operators will apply equally to the equations described below.

As shown in FIG. 4, each of the memory addresses add indicates two codewords, and LEN and VAL are determined by the index idx. The encoding of Coeff_token with reference to the memory structure for the VLCT according to the first embodiment of the present invention includes the following steps. For example, generating an address (add) and an index (idx) from the input Tc and T1s values, reading LEN and VAL from LEN_M and VAL_M according to the generated address (add), and generating the index determining LEN and VAL according to (idx).

An operation of encoding Coeff_token in the memory structure for the VLCT according to the first embodiment of the present invention will be exemplified. When Tc = 2 and T1s = 1 are input, parameters (cn = 5, j = 0), an address (add = 2) and an index (idx = 1) are generated by the equations (1) through (5). Referring to the memory structure shown in Fig. 4, LEN = 6 and VAL = 4 are determined. That is, a bit string '000100' having a bit stream length of '6' and a bit stream value of '4' is determined. The determined bit stream '000100' is the same as the bit stream '000100' when Tc = 2 and T1s = 1 in VLCT0 of FIG. 3A.

5 is a block diagram illustrating an entropy encoder according to an embodiment of the present invention.

Referring to FIG. 5, the entropy encoder 160 includes a codenum generator 161, a first address and index generator 162, a first storage 163, A second address and index generator 164, a second storage 165, a base (hereinafter referred to as a base) calculator 166, a bit stream generator 167, and an adder 168. ).

Each of the codenum generator 161 and the base calculator 166 generates a plurality of parameters required for encoding Coeff_token according to the second embodiment of the present invention. The first address and index generator 162 generates a first address add for accessing the first storage unit 163 and a first index for determining the read VAL according to the first address add. Create (idx). The second address and index generator 164 generates a second address Add for accessing the second storage unit 165 and a second index for determining the read LEN according to the second address Add. Create (Idx).

According to an embodiment of the present invention, the codewords present in the VLCTs VLCT0 to VLCT2 are divided into LEN and VAL. LEN and VAL are stored in their respective storage. That is, the LEN is stored in the second storage unit 165 for storing bit stream length information of the codeword. The VAL is stored in the first storage unit 163 for storing bit stream value information of the codeword. According to the second embodiment of the present invention, the size of the bit stream length (LEN) information of the codeword is reduced and stored in the second storage unit 165.

FIG. 6 is a diagram illustrating a correlation between bit stream length (LEN) information of a codeword and a parameter codenum according to an embodiment of the present invention.

Referring to Equations 1 to 3, a parameter codenum is generated according to Tc and T1s and lists the codewords of VLCT shown in FIGS. 3A to 3C in one dimension. If LEN and VAL are listed in one dimension by applying a parameter (codenum), there is a correlation between LEN and a parameter (codenum). FIG. 6 is a diagram for clearly identifying such correlations, in which LENs are arranged according to a parameter (codenum).

Referring to FIG. 6, as the parameter codenum increases, the LEN also increases. In addition, when the parameter codenum is grouped around a parameter codenum that is a multiple of 8, the sizes of the LENs included in each of the groups change slowly. For example, the LEN of VLCT0 included between the parameter codenum8 and the parameter codenum15 has a value of 7-9. If 7 is set as the default value in this interval, the LEN in this interval may be represented by any one of values of 0, 1, or 2. In other words, LEN can be expressed only with a default value and 2 bits. In FIG. 6, values in parentheses indicate default values of respective sections. According to the default values shown, all corresponding LENs have a value between 0 and 3.

7 is a diagram illustrating a process of setting a default value for a parameter (codenum) according to an embodiment of the present invention.

As mentioned above, Tc and T1s are independent variables that can be utilized in Coeff_token encoding. That is, a default value can be generated from Tc and T1s. Referring to FIG. 6, the default values of the groups grouped according to the parameter codenum increase by 1 or 2 as the parameter codenum increases. Therefore, as shown in Equations 6 to 10, the default value may be modeled as a function of a codenum.

7 shows values for the variables of Equations 6 to 10 for the parameter codenum. For example, comparing the default value base0 calculated from Equations 11 to 13 with the default value illustrated in VLCT0 of FIG. 6 has the same default value.

Equations 14 to 16 are equations for generating default values for the VLCTs VLCT0 to VLCT2 illustrated in FIGS. 3A to 3C, respectively. That is, Equation 14 is an equation for generating a default value for VLCT0 shown in FIG. 3A, Equation 15 is an equation for generating a default value for VLCT1 shown in FIG. 3B, and Equation 16 is shown in FIG. 3C. A formula for generating a default value for the illustrated VLCT2.

8 is a diagram illustrating a memory structure for a variable length code table according to a second embodiment of the present invention.

The bit stream length LEN of the codewords of FIGS. 3A to 3C is expressed by Equation 17 according to a default value calculated by Equations 14 to 16. The bit stream length LEN of the codeword is generated by the adder 168.

The parameter (offset) of equation (17) represents the difference between the LEN and the default value of the VLCT. In other words, the offset is an increase value that should be increased based on the calculated default value in order to generate the LEN. Such an offset is stored in the second storage unit (see 165 of FIG. 5) according to the second embodiment of the present invention. Since the offset ranges from 0 to 3, two bits of storage space are allocated. On the other hand, in a general memory device (eg, the second storage unit 165), data is stored and referenced in units of 8 bits. Therefore, one memory address must be assigned to four parameters.

Equations 18 to 19 are equations for generating a second memory address Add in the memory structure according to the second embodiment of the present invention. Equation 20 is an equation for generating a second index Idx for identifying four parameters (offsets) included in the same second memory address Add.

An operation of encoding Coeff_token in the memory structure for the VLCT according to the second embodiment of the present invention will be described with reference to FIGS. 5 and 8. As an example, the bit stream length LEN of a codeword will be calculated. When Tc = 5 and T1s = 2 are input, the codenum generator 161 generates a parameter codenum value. A parameter codenum value is generated by the equations (1) through (3). The generated parameter value is 14. The second address and index generator 164 generates a parameter value or a second address and index value. That is, the parameter ad = 3 is generated by the equation (6), and the parameter sa = 0 is generated by the equation (18). In addition, a second address (Add = 3) is generated by Equation 19, and a second index (Idx = 2) is generated by Equation (20).

A parameter stored in the second storage unit 165 is read according to the generated second address Add according to the second index Idx. The read parameter value is 3. The base generator 166 generates a default value. Equations 7 to 16 generate default values (base0 = 7, base = 4). The adder 168 adds the generated default value and the read offset value. That is, the LEN value is generated by adding the default value (base = 4) and the parameter (offset = 3) value generated by Equation 17. The generated LEN is 7, which is the same as the bit stream length of the codeword when Tc = 5 and T1s = 2 in VLCT1 of FIG. 3B.

9 is a flowchart illustrating an operation of encoding Coeff_token in a memory structure according to a second embodiment of the present invention.

5 and 9, an operation of encoding Coeff_token with reference to a memory structure for a VLCT according to the second embodiment includes the following steps. The codenum generator 161 generates a parameter codenum from the input Tc and T1s (step S110). The first address and index generator 162 generates a first address (add) and a first index (idx) value for accessing the first storage unit 163 with reference to the generated parameter codenum value ( Step S120). The bit stream value VAL of the codeword is read from the first storage unit 163 according to the generated first address add and the first index idx (step S130).

The second address and index generator 163 generates a second address (Add) and a second index (Idx) value for accessing the second storage unit 165 with reference to the generated parameter codenum value ( Step S140). The offset value is read from the second storage unit 165 according to the generated second address Add and the value of the second index Idx (S150). The base generator 166 generates a default value (step S160). The bit stream length LEN of the codeword is generated with reference to the generated default value and the offset value (step S170). The generated LEN and VAL are transmitted to the bit stream generator 167 to generate a bit stream (step S180).

According to the second embodiment of the present invention, codewords indicated by the VLCT are divided into length information and value information. Length information and value information of the codewords are stored in each allocated memory area. The size of the length information of the codeword is configured to be reduced, and the length information of the reduced codeword is stored in the allocated memory area. Thus, the operating speed of the encoder is improved and the memory area in which the VLCT is stored is reduced.

100: encoder
110: motion estimation unit
120: motion compensation unit
130: intra prediction unit
140: filter
150: rearrangement
160: entropy encoder
161: code number generation unit
162: first address and index generator
163: first storage unit
164: second address and index generator
165: second storage unit
166: base calculation unit
167: bit stream generation unit
168: adding up

Claims (15)

delete In the context adaptive variable length code method of encoding:
Generating a parameter value codenum with reference to information on the number of non-zero coefficients Tc of the quantized transform coefficients and the number of coefficients T1s having a magnitude of 1;
Calculating a default value of a bit stream length of a codeword with reference to the parameter value (codenum);
Reading an offset value from a storage unit according to an address (Add) and an index (Idx) value generated by referring to the parameter value codenum;
Generating bit stream length information of a codeword according to the base and the offset; And
Generating a bit stream of a codeword using bit stream value information of a codeword and bit stream length information of the generated codeword,
The parameter value codenum is

The encoding method defined by. The method of claim 2,
The variables (j, cn),

; And

The encoding method defined by. In the context adaptive variable length code method of encoding:
Generating a parameter value codenum with reference to information on the number of non-zero coefficients Tc of the quantized transform coefficients and the number of coefficients T1s having a magnitude of 1;
Calculating a default value of a bit stream length of a codeword with reference to the parameter value (codenum);
Reading an offset value from a storage unit according to an address (Add) and an index (Idx) value generated by referring to the parameter value codenum;
Generating bit stream length information of a codeword according to the base and the offset; And
Generating a bit stream of a codeword using bit stream value information of a codeword and bit stream length information of the generated codeword,
The base is composed of a plurality of default values, each encoding method corresponding to a plurality of variable length code table. The method of claim 4, wherein
The plurality of default values,

The encoding method defined by. The method of claim 5, wherein
The variables base0, k2, and sel are referenced to the generated parameter codenum value.

;

;

;

;

;

;

; And

The encoding method defined by. In the context adaptive variable length code method of encoding:
Generating a parameter value codenum with reference to information on the number of non-zero coefficients Tc of the quantized transform coefficients and the number of coefficients T1s having a magnitude of 1;
Calculating a default value of a bit stream length of a codeword with reference to the parameter value (codenum);
Reading an offset value from a storage unit according to an address (Add) and an index (Idx) value generated by referring to the parameter value codenum;
Generating bit stream length information of a codeword according to the base and the offset; And
Generating a bit stream of a codeword using bit stream value information of a codeword and bit stream length information of the generated codeword,
The address (Add) and the index (Idx) with reference to the generated parameter (codenum) value,

; And

The encoding method defined by. The method of claim 7, wherein
The variables (sa, ad) with reference to the generated parameter (codenum) value,

; And

The encoding method defined by. In the context adaptive variable length code method of encoding:
Generating a parameter value codenum with reference to information on the number of non-zero coefficients Tc of the quantized transform coefficients and the number of coefficients T1s having a magnitude of 1;
Calculating a default value of a bit stream length of a codeword with reference to the parameter value (codenum);
Reading an offset value from a storage unit according to an address (Add) and an index (Idx) value generated by referring to the parameter value codenum;
Generating bit stream length information of a codeword according to the base and the offset; And
Generating a bit stream of a codeword using bit stream value information of a codeword and bit stream length information of the generated codeword,
The bit stream length information of the codeword is generated by adding the base and the offset. A codenum generator for generating a parameter from the quantized transform coefficients;
A default value calculator for calculating a default value according to the generated parameter value;
A storage unit for storing a bit stream length information value of a variable length codeword and a difference value of the default value;
An adder configured to generate bit stream length information of the variable length codeword by adding the calculated default value and a difference value read according to the generated parameter value;
A storage unit for storing bit stream value information of the variable length codeword; And,
And a bit stream generator for generating a bit stream of a codeword according to the bit stream length information of the generated variable length codeword and the bit stream value information of the variable length codeword read out according to the generated parameter value.
And a storage unit for storing the increment value and a storage unit for storing the bit stream value of the variable length codeword. 11. The method of claim 10,
And the difference value comprises two bits. 11. The method of claim 10,
And an address and index generator for generating an address and an index for accessing the storage according to the generated parameter value. delete delete delete

Images