KR100801967B1 - Encoder and decoder for Context-based Adaptive Variable Length Coding, methods for encoding and decoding the same, and a moving picture transmission system using the same - Google Patents

Abstract

The present invention relates to an encoder for encoding image data consisting of frames, a decoder for decoding, and an encoding method and a decoding method therefor. In the context-based adaptive variable length coding method (CAVLC) according to the present invention, the number of trailing ones T1s and the number of nonzero coefficients in a series of transform coefficients for the target block of the frame Num Coeff); Determining a variable length coding table from the number NumCoeff of the nonzero coefficients in the neighboring block of the target block and the number NumCoeff of the nonzero coefficients in the block at the same position as the target block in the previous frame; And obtaining a code value for the target block from the number T1s of trailing 1 and the number NumCoeff of the nonzero coefficients in the selected variable length coding table and the series of transform coefficients. The context-based adaptive variable length coding method according to the present invention can perform encoding more efficiently.

CAVLC, H.264, MPEG-4 AVC, Variable Length Coding Table

Description

Context-Based Adaptive Variable Length Coding Encoder and Decoder, Context-Based Adaptive Variable Length Coding Method and Decoding Method, and Video Transmission System Using the Same. {Encoder and decoder for Context-based Adaptive Variable Length Coding, methods for encoding and decoding the same, and a moving picture transmission system using the same}

1 is a diagram schematically illustrating a video transmission and reception system that is generally used.

2 is a schematic block diagram illustrating a process of performing encoding in a conventional H.264 encoder.

FIG. 3 is a diagram illustrating a zigzag scanning sequence of a 4 × 4 block for encoding in an H.264 encoder.

4 is a block diagram schematically illustrating a process of performing decoding in a decoder for reconstructing a video by decoding a bitstream generated by an H.264 encoder.

5 is a diagram illustrating an arrangement relationship of blocks constituting one frame.

6 shows an H matrix used for performing frequency conversion in the H.264 scheme.

7 is a view showing the position of a block in the n-k th frame and n th frame in the present invention.

8 is a flowchart illustrating a process of encoding a block in the context-based adaptive variable length coding method according to the present invention.

9 is a flowchart illustrating in detail the steps of encoding the number of trailing 1 and the number of non-zero coefficients in the context-based adaptive variable length encoding method according to the present invention.

10 is a flowchart showing in detail a step of determining a variable length coding table in the context-based adaptive variable length coding method according to the present invention.

11 is a flowchart illustrating a step of decoding data in a decoded bitstream by the context-based adaptive variable length coding method according to the present invention.

12 is a diagram illustrating a relationship between a target block, neighboring blocks, and blocks in a previous frame, and encoding information in these blocks in an exemplary embodiment of the present invention.

<Description of the symbols for the main parts of the drawings>

12, 14: image processing unit 20: block in the frame

16, 18: monitor 30: encoder

31: n-th frame storage unit 32: summer

34: frequency converter 36: quantization unit

38: reordering unit 40: entropy coding unit

42: inverse quantization unit 44: inverse transform unit

46: summer 50: selection in frame

52: n-1th frame storage unit

54: motion compensation unit 56: motion estimation unit

60: filter unit 62: reconstructed n-th frame storage unit

100: decoder 110: entropy decoding unit

112: rearrangement unit 114: inverse quantization unit

116: inverse transform unit 118: summer

120: filter 122: reconstructed frame storage unit

124: intra frame prediction unit 126: motion compensation unit

128: comparison frame storage

208: upper block of the target block 210: left block of the target block

212: target block

The present invention relates to an encoder for decoding image data consisting of frames, a decoder for decoding, and an encoding method and decoding method for the same. Specifically, an encoder, a decoder for using context-based adaptive variable length coding (CAVLC), and The present invention relates to an encoding method and a decoding method. More specifically, when context-based adaptive variable length encoding of a block constituting a frame, an encoder using encoding information of a block at the same position as the block encoded in the previous frame and a block adjacent to the block encoded in the current frame; A decoder, an encoding method and a decoding method for the same.

Digital devices such as personal computers, wireless devices, surveillance devices, video conferencing systems, and digital broadcasting set-top boxes transmit and receive multimedia data including video data. Therefore, the efficient transmission and reception of image data in such digital equipment is becoming increasingly important.

1 shows an example of a video transmission system. The terminal A side is composed of the monitor 16 and the image processing unit 12, and the terminal B side is also composed of the monitor 18 and the image processing units 12 and 14. The image displayed on the monitor 18 of the A side terminal is compressed and encoded by the image processing unit 12 and transmitted to the image processing unit 14 of the B side terminal through the Internet 10. At this time, one block 20 of the video displayed on the monitor 16 of the A side terminal may be the same as the block 20 shown on the A side terminal according to the encoder of the image processor 14 and the decoder of the image processor 14. And even coarser. In the B side terminal, the reverse process may be performed to transmit the video to the A side terminal.

Various image compression techniques are currently used to remove the temporal and spatial redundancy of data in the frames of successive images. In this way, a method of removing redundancy of data in one frame and a method of removing redundancy of data over several frames are used.

As an example of how to remove redundancy of data within one frame, when encoding a target block in a frame composed of multiple blocks, the target block is expected from a previously encoded block, and the expected block and the target block are encoded. There is a way to encode only the difference from. The method of predicting the target block assumes that each pixel moves in the same direction and size in each frame.

A block (hereinafter, referred to as a 'difference value block') for a difference value between the expected block and the actual target block thus obtained is converted into a frequency domain. A frequently used discrete cosine transform (DCT) for this frequency domain is the Discrete Cosine Transform (DCT). For the same purpose, standards such as H.264 use matrices of discrete cosine transforms transformed into integers. This conversion to the frequency domain primarily compresses the block to the difference.

The transformed block is then quantized in a specific quantization step. The larger the size of the quantization step, the larger the quantization error. Therefore, the encoder calculates an appropriate quantization step to quantize the transformed block. In this way, compression is also performed in the quantization process, but this is lossy compression.

The quantized block is scanned to form a bit string of transform coefficients, and then the bit string is bitstreamed by an entropy coder. In this process, the quantized block is further compressed. The most frequently used method of entropy encoder is variable length coding in which the most frequently occurring code words are short bit strings and the rarely occurring code words are long bit strings.

As a technique for compressing an image by following the above-described steps, there is H.264, or MPEG-4 AVC. Looking at the process of image compression in H.264 in relation to Figure 2 as follows. 2 is a block diagram of an H.264 encoder 30. In the encoder 30, the n th frame of the video is input from the n th frame storage unit 31 to the encoder 30. This frame 31 is processed in units of macroblocks of size 16x16 pixels. Each macroblock is encoded by one of an intra mode and an inter frame mode. Either mode is used or the expected macroblock P is generated. In the intra-frame mode, the current frame Fn is processed by the transform unit 34 and the quantizer 36, and then the frame uF ′ reconstructed by the inverse quantizer 42 and the inverse transform unit 44. n) and macroblock P from it. In interframe mode, the macroblock P predicts the motion of the current frame in the motion estimation section 56 based on one or more comparison frames F ′ _{n −1} and the current frame Fn, and compensates for the motion. It is generated by compensating for the movement of the current frame in the motion compensation unit 54. Whether to use the interframe mode or the intraframe mode is determined by the intraframe prediction selector 50.

As shown in FIG. 2, the macroblock P thus predicted and the macroblock of the current frame Fn are input to the subtractor 32 to obtain a difference macroblock Dn. Subsequently, the difference macroblock Dn is DCT-converted by the frequency domain transform unit 34 and then quantized in a quantization step Qstep by the quantization unit 36.

The quantized macroblock is scanned in a zigzag manner in the order shown in FIG. 3 by the reordering unit 38 so that the quantized transform coefficients are arranged in order. Then, the aligned series of transform coefficients are encoded by the entropy encoder 40 and then output in the form of a bitstream.

Meanwhile, the frame uF'n reconstructed by the inverse transformer 44 and the inverse quantizer 42 in the encoder is stored as the reconstructed frame F'n after passing through the filter 60 to encode another frame. Can be used later. The filter 60 is a deblocking filter for mitigating distortion occurring between macroblocks of the reconstructed frame uF'n.

4 illustrates a process of reconstructing a video by decoding a bitstream generated by an H.264 encoder. When the bitstream is input to the decoder 100, the entropy decoding unit 110 reconstructs the transform coefficient of the macroblock. The reconstruction coefficient is reconstructed in the form of a macroblock in the reordering unit 112. The reconstructed macroblock in the decoder 100 generates a difference value macroblock D'n through the inverse quantization unit 114 and the inverse transform unit 116. This difference macroblock D'n and the expected macroblock P are combined in summer 118 to produce a reconstructed macroblock uF'n. The reconstructed macroblock uF'n is used to generate the expected macroblock P in the in-frame mode. In addition, the macroblock uF'n generates a frame F'n through a deblocking filter.

As the entropy encoding method used by the entropy encoder 40 and the entropy decoder 110 described above, there is a context-based adaptive variable length coding (CAVLC) scheme. In this method, one of a plurality of Variable Length Coding Tables is selected based on encoding information of an already encoded block. For example, in H.264, variable length coding tables such as Num-VLC0, Num-VLC1, Num-VLC2, and FLC are used. In the H.264 method, when encoding a target block, one of four variable length coding tables is selected from the four variable length coding tables based on the number of non-zero coefficients (NumCoeff) among the encoding information of the left and upper neighboring blocks that are already encoded. The number of trailing ones (T1s) and the number of nonzero coefficients (NumCoeff) are encoded.

A conventional technique, for example, a method of selecting one of variable length coding tables in H.264 will be described in detail as follows. When encoding a frame, since the upper left block is usually encoded in the lower right direction, the encoding information of these blocks can be used since the left or upper block is already encoded when the target block is encoded. 5 is a diagram illustrating an arrangement relationship of blocks constituting one frame. When the target block 212 is to be encoded, the block 210 on the left side and the block 208 on the upper side are often encoded. In H.264, when the number of nonzero coefficients in the upper block 208 is N _U , and the number of nonzero coefficients in the left block 210 is N _L , the nonzero expected in the target block 212 is The number N of coefficients is round (N _L + N _U ) / 2. In H.264, one of four variable length coding tables is selected as shown in Table 1 below based on the number N of nonzero coefficients in the expected target block 212.

N Variable Length Coding Table 0, 1 Num-VLC0 2, 3 Num-VLC1 4, 5, 6, 7 Num-VLC2 8 or more FLC

In H.264, in the variable length coding table defined in this way, the number of nonzero coefficients (NumCoeff) and the number of trailing ones (T1s) T1s in the actual target block 212 are referred to the target block. Get the first partial code word of the bitstream for.

H.264 then encodes the sign of trailing 1, encodes the magnitude of the nonzero coefficient except trailing 1, encodes the number of zeros that precede the last nonzero coefficient, and finally Encode the location.

In conventional H.264, at least one of encoding information N _L or N _U for the left or right block of the target block may not be obtained at the time of encoding the target block. For example, the target block and the adjacent block belong to different slices, or the target block is located at the left end or top end of the frame. If N _L is not obtained, the value of N is set to N _U. On the other hand, when N _U cannot be obtained, the value of N is set to N _L. Finally, if neither N _U nor N _L can be obtained, the value of N is set to zero.

However, when entropy encoding using context-based adaptive variable-length coding in video-related standards, including H.264, objects within the current frame to predict the number of non-zero coefficients (N) in the block to be encoded. Since only non-zero coefficients (N _U , N _L ) are considered in blocks located above or to the left of the block, there is a problem in that the hit ratio for the number N of non-zero coefficients expected to be in the target block is not high. Therefore, the conventional video related standard does not use the optimal variable length coding table when context-based adaptive variable length coding, there is a problem that the efficiency of entropy coding is inferior.

The present invention provides an encoding method and decoding method for selecting a variable length encoding table that is more efficient when context-based adaptive variable length encoding of data including a video, an encoder and a decoder using the method, and a video transmission system using the same. It aims to do it.

In addition, the present invention can select a more efficient variable length encoding table in consideration of encoding information of a previous frame as well as encoding information of a neighboring block of a target block of a frame to be encoded when context-based adaptive variable length encoding of data including a video is performed. The purpose is to provide a way.

In addition, the present invention provides a method for compressing and transmitting the same data with a smaller number of bits by further improving the efficiency of context-based adaptive variable length coding, a method of decoding such data, an encoder and decoder using the same, and a video transmission system. It aims to provide.

Principles of the Invention

The present invention uses the principle that when context-based adaptive variable length coding of one block of a video frame is not only highly correlated between the target block and the adjacent frame, it is also highly correlated with the block of the same or adjacent position in the previous frame. That is, the present invention not only considers encoding information of a block in a current frame when encoding a block of a current frame of a video, but also considers encoding information of a block located at the same or adjacent position as a block to be encoded in a previous frame. The encoding of the target block is performed. Here, the positions of the blocks of the current frame and the previous frame refer to the physical positions of the blocks in each frame without considering the motion vectors of the blocks.

This method of encoding is particularly effective even when encoding information for an adjacent block of the encoding target block of the current frame cannot be obtained. That is, when the neighboring block of the encoding target block belongs to another slice and thus the encoding information on the neighboring block cannot be accessed or the neighboring block available at the edge of the frame is limited, the previous frame having a high correlation with the target block to be encoded By using blocks of, it is possible to perform context-based adaptive variable length coding more efficiently.

Encoder and Encoding Method According to the Invention

Looking specifically at an encoder according to the present invention using such a principle, the encoder according to the present invention performs context-based adaptive variable length coding (CAVLC) on a quantized transform coefficient of data including a first frame and a subsequent second frame. do. The encoder according to the present invention includes a previous frame encoding information storage unit and a context-based adaptive variable length encoder.

In the description and claims of this specification, "following" means later in time, but not necessarily immediately after. That is, the meaning that the second frame follows the first frame means that when encoding or decoding, the first frame is encoded or decoded before the second frame. Thus, the first frame of the movie N ₁ th frame in the encoder according to the invention, when said second frame is N _2-th frame, is established a relationship of N ₁ <N _2. Preferably, N ₁ frame is N ₂ A reference frame used to generate the above-described P frame when decoding a frame. Of course, the relationship of N ₁ + 1 = N ₂ may be satisfied. That is, the second frame may be a frame immediately after the first frame.

Likewise, in the description and claims of this specification, a "previous" frame means temporally ahead of the target frame, but not necessarily the immediately preceding frame. That is, in the description and claims herein, "previous frame" refers to all frames that are encoded or decoded prior to the target frame that temporally precedes the target frame.

The previous frame encoding information storage unit stores encoding information obtained while context-based adaptive variable length encoding of the first frame. The encoded information storage unit may be configured with any data storage device for storing encoded information of the first frame, such as a normal RAM or a high speed cache. Preferably, the encoding information of the first frame may be stored in a previous frame encoding information storage unit in the form of a matrix.

The encoding information refers to information necessary in the process of encoding a block. Specifically, in context-based adaptive variable length coding to which the present invention is applied, encoding information includes (i) the number of nonzero coefficients in a block, (ii) the number of trailing 1 in the block, and (iii) the sign of trailing 1 , And (iv) the total number of zeros before the nonzero coefficient. In the present invention, one or more of such encoded information is used. Preferably, the encoder according to the present invention uses the number NumCoeff of the non-zero coefficient in the block of the encoding information.

In addition, the context-based adaptive variable length encoding unit context-based adaptive variable length encoding of the second frame in consideration of the encoding information stored in the previous frame encoding information storage unit. Preferably, the context-based adaptive variable length encoder generates a reference value in consideration of the stored encoding information, and performs encoding based on the reference value. The context-based adaptive variable length encoder may perform context-based adaptive variable length coding in consideration of adjacent blocks of a target block to be encoded in the current frame.

Preferably, the encoder includes a block divider for dividing each frame into a plurality of blocks; A frequency converter for converting the data value of the divided block into a frequency domain; And a quantization unit for quantizing data values of the block transformed into the frequency domain.

The block divider can preferably be divided into blocks of a suitable size using methods already known in the art. The block divider may divide each frame into blocks having a constant 4x4 size, and may variably size the blocks to further improve compression efficiency. In general, when a variable block size is used, a large portion of the frame is divided into smaller blocks, and a small portion of the frame is divided into larger blocks.

The frequency converter of the encoder converts the information in the time domain into the frequency domain. Transforms used for this purpose include a Fourier transform, a discrete cosine transform (DCT), and the like. In an encoder such as the present invention, these frequency transforms may be converted to exact figures, but preferably, proximity values may be used to increase the computation speed and reduce the computation amount. For example, H.264 uses an H matrix as shown in FIG. 6, which simplifies the matrix of discrete cosine transform into an integer form. Since the inverse of the integer frequency conversion matrix is also expressed as an integer, the problem that the inversely converted value after the conversion is inconsistent with the original value can be eliminated.

The quantization unit can quantize the transform coefficients with a predetermined quantization step size. Smaller quantization steps have the advantage of reducing quantization errors by quantizing the transform coefficients more accurately, while increasing the size of the quantized data. Accordingly, the quantization unit preferably determines quantization steps of appropriate size in consideration of these advantages and disadvantages. For example, the quantization step (Qstep) used for quantization in H.264 is shown in Table 2 below. For each quantization step Qstep, a quantization parameter QP of 0 to 51 is provided. Each quantization step is increased by about 12%.

Preferably, the encoder performs transformation, quantization, and encoding on a block basis of each frame. As described above, each block may be a block of 4x4 size uniformly, or blocks of various sizes may be used according to the structure of the frame.

Preferably, the encoding information is the number NumCoeff of nonzero coefficients in a block that is already encoded. For example, in a 4x4 pixel block, such as Table 3 below, the nonzero coefficients are 3, -1, -1, 1, 1, so the number of nonzero coefficients (NumCoeff) in this block is 5. .

Preferably, the number NumCoeff of the non-zero coefficients of the block of the first frame of frequency transform and quantization is used when encoding the block of the second frame after the previous frame encoding information storage unit is stored. More preferably, in the present invention, the number of non-zero coefficients NumCoeff in the block at the same position as the block to be encoded in the second frame among the blocks of the first frame is determined by the block adjacent to the block to be encoded in the second frame. The reference value N is determined in consideration with the number of nonzero coefficients NumCoeff.

Preferably, the reference value (N) is a block adjacent to the block to be encoded in the second frame and the number N of nonzero coefficients in the block at the same position as the block to be encoded in the second frame among the blocks of the first frame. NumCoeff is the number of nonzero coefficients expected to be present in the target block to be encoded of the second frame, derived from the number of nonzero coefficients of.

Preferably, the adjacent block of the block to be encoded of the second frame is a block that is encoded before the block to be encoded in consideration of the order of encoding the blocks of the second frame. For example, in the case of H.264, the left block and the upper block of the target block correspond to adjacent blocks.

A relationship between a reference value N, encoded information N _L , encoded information N _U, and encoded information N _P will be described with reference to FIG. 7. N _L is the number of nonzero coefficients (NumCoeff) in the block located to the left of the target frame in the same frame as the target block (nth frame), and N _U is the number of target blocks in the same frame as the target block (nth frame) The number of nonzero coefficients (NumCoeff) in the upper block, N _P is the number of nonzero coefficients (NumCoeff) in the block at the same position as the target block in the previous frame (nk th frame, where k ≥ 1) to be. Preferably, the reference value N, that is, the number N of non-zero coefficients expected in the encoding target block is determined by the following equation.

N = round {αx (N _U + N _L ) + βxN _P }

However, α and β are positive real numbers that satisfy the relationship of 2α + β = 1.

In some cases, the encoding information for the adjacent frame of the target block cannot be obtained. The bitstream output from the encoder according to the present invention may be configured in units of slices. A slice is a unit that organizes a series of blocks so that a frame can be scanned at a faster speed when a variable size block is not used. Thus, one frame is composed of a plurality of slices. One slice is itself encoded and decoded into one unit, and when encoding or decoding a block within that slice, only one piece of information within that slice is stored without any information in the other slice, except for some parameters for the deblocking filter. A block within a slice can be encoded or decoded. Therefore, when the target block and the adjacent frame are located in different slices, encoding information of the adjacent frame may not be obtained when decoding the target block. In addition, when the target block is located at the edge of the frame, there may be no adjacent frame at all.

For this reason, when N _U cannot be obtained at the time of encoding the target block, the reference value, that is, the number N of non-zero coefficients expected in the encoding target block is obtained as follows.

N = round (αxN _L + βxN _P )

Where N _L is the number of nonzero coefficients in the block located to the left of the target block, N _P is the number of nonzero coefficients in the block at the same position as the target block in the previous frame, and α and β are 2α + It is a positive real number that satisfies the relationship of β = 1.

On the contrary, when N _L is not obtained at the time of encoding the target block, the reference value, that is, the number N of non-zero coefficients expected in the encoding target block is obtained as follows.

N = round (αxN _U + βxN _P )

Where N _U is the number of nonzero coefficients in the block located above the target block, N _P is the number of nonzero coefficients in the block at the same position as the target block in the previous frame, and α and β are 2α + It is a positive real number that satisfies the relationship of β = 1.

If N _U and N _L are not obtained at the time of encoding the target block, the reference value, that is, the number N of non-zero coefficients expected in the encoding target block is determined as Np.

Preferably, in the present invention, the entropy encoder of the encoder according to the present invention selects one of a plurality of variable encoding tables and performs encoding based on the N value. For example, when the present invention is applied to an H.264 encoder, one of four reference tables, Num-VLC0, Num-VLC1, Num-VLC2, and FLC, can be selected and encoded based on the N value. have. Selection of the variable coding table according to the N value may use the method described in Table 1 as it is or may be modified.

The variable coding tables Num-VLC0, Num-VLC1 and Num-VLC2 may be given as Tables 4, 5 and 6, respectively. Where NumCoeff is the number of nonzero coefficients in the block, and T1s is the number of trailing ones.

Numcoeff Of T1s 0 One 2 3 0 One - - - One 000011 01 - - 2 00000111 0001001 001 - 3 000001001 00000110 0001000 00011 4 000001000 000001011 000000101 000010 5 0000000111 000001010 000000100 0001011 6 00000000111 0000000110 0000001101 00010101 7 000000001001 00000000110 0000001100 00010100 8 000000001000 00000001001 000000001010 000000111 9 0000000000111 000000001011 000000000101 0000000101 10 0000000000110 0000000001101 0000000001111 00000001000 11 00000000000011 0000000001100 0000000001110 000000000100 12 00000000000010 00000000000100 00000000000110 0000000000101 13 00000000000101 00000000000111 000000000010001 00000000001001 14 000000000000011 000000000000010 000000000010000 0000000000000011 15 00000000000000001 00000000000000011 00000000000000010 00000000000000101 16 00000000000000000 000000000000001001 0000000000000010001 0000000000000010000

Numcoeff Of T1s 0 One 2 3 0 11 - - - One 000011 011 - - 2 000010 00011 010 - 3 001001 001000 001010 101 4 1000001 001011 100101 0011 5 00000111 1000000 1000010 00010 6 00000110 1000011 1001101 10001 7 000001001 10011101 10011100 100100 8 000001000 000001011 000000101 1001100 9 0000000111 000001010 000000100 10011111 10 0000000110 0000001101 0000001100 10011110 11 00000000101 00000000111 00000001001 000000111 12 00000000100 00000000110 00000001000 0000000101 13 000000000011 000000000010 000000000100 000000000111 14 0000000000011 000000000101 0000000000010 0000000001101 15 00000000000001 00000000000000 000000000000111 0000000001100 16 000000000000101 000000000000100 0000000000001101 0000000000001100

Numcoeff Of T1s 0 One 2 3 0 0011 - - - One 0000011 0010 - - 2 0000010 101110 1101 - 3 000011 101001 010110 1100 4 000010 101000 010001 1111 5 101101 101011 010000 1110 6 101100 101010 010011 1001 7 101111 010101 010010 1000 8 0110101 010100 011101 00011 9 0110100 010111 011100 00010 10 0110111 0110110 0110000 011111 11 01111001 0110001 01111010 0110011 12 01111000 01111011 01100101 01100100 13 000000011 000000010 000000100 000000111 14 0000000011 000000101 0000001101 0000001100 15 0000000010 00000000011 00000000010 00000000001 16 0000000000001 000000000001 00000000000001 00000000000000

FLC can be given in the form of 6-bit xxxxyy. Where xxxx is a binary representation of the number of non-zero coefficients (NumCoeff) minus 1, and yy represents the number of trailing ones. In this case, when the number NumCoeff of the non-zero coefficient is 0, the 000011 code value is used.

As described above, when calculating N values by applying equations 1 to 3, the values of α and β vary depending on the size of the quantization step (Qstep) used when the quantization unit quantizes the data values of the block. It is preferable. More preferably, the quantization unit uses values of α and β as shown in Table 7 below for the quantization step (or for the quantization parameter QP).

QP Qstep α β 0-16 0.625-4 0.125 0.75 17-28 4.5-16 0.145 0.71 29-51 18-224 0.165 0.67

The encoder according to the present invention may further include an input frame memory, a reference frame memory, a motion estimation and compensation unit, a subtractor, a transformer, a quantizer, and an encoder. The input frame storage unit stores data of a target frame input to the encoder. The reference frame storage unit stores data of a previous frame of the target frame. The motion estimation and compensation unit selects a block of the previous frame closest to the block of the target frame. The subtractor obtains a difference block between the block of the previous frame and the block of the target frame obtained by the motion estimation and compensation unit. Subsequently, the converter converts the difference block obtained in the subtractor into a frequency domain. The subsequent quantization unit quantizes the block transformed into the frequency domain with a predetermined quantum step size (Qstep). Finally, the encoder performs context-based adaptive variable length coding (CAVLC) on the quantized block in the quantizer.

Hereinafter, a method of context-based adaptive variable length encoding of video data consisting of a plurality of frames will be described. As shown in FIG. 8, in the general context-based adaptive variable length encoding method, encoding the number of trailing ones (T1s) and the number of nonzero coefficients (NumCoeff) in a series of transform coefficients ( 252); Encoding (254) the sign (plus or minus) of each trailing 1; Encoding 256 the magnitude of each non-zero coefficient; Encoding (258) a number of coefficients of zero in front of the last non-zero coefficient in the series of transform coefficients; And encoding (260) the positions of the non-zero coefficients.

In the context-based adaptive variable length coding method according to the present invention, the step 252 of encoding the number T1s of trailing 1 and the number NumCoeff of nonzero coefficients is performed by following the detailed steps as shown in FIG. 9. do. First, the number of trailing ones (T1s) and the number of nonzero coefficients (NumCoeff) are determined in the series of transform coefficients (282). Here, the number of nonzero coefficients is determined from the number of nonzero coefficients NumCoeff in the adjacent block of the target block and the number of nonzero coefficients NumCoeff in the block at the same position as the target block in the previous frame. . In operation 284, a variable length encoding table for encoding the block is determined. Finally, a code word for the target block is obtained from the number of trailing ones and the number of nonzero coefficients in the selected variable length coding table and the series of transform coefficients (286).

Preferably, in the context-based adaptive variable length coding method according to the present invention, the neighboring block is a block coded before the target block in the order of encoding the blocks of the frame. More preferably, the adjacent block is the left block and the upper block of the target block.

Determining the variable length coding table 284 includes: obtaining (302) the number N _L of nonzero coefficients in the left block of the target block as shown in FIG. Obtaining (304) the number N _U of nonzero coefficients in the upper block of the target block; Obtaining the number N _P of non-zero coefficients in the block in the same location as the current block in the previous frame (306); Calculating a reference value N (308); And selecting 310 a variable length encoding table to be used based on the value of N.

In operation 308 of calculating the reference value N, Equations 1 to 3 or N = N _P are set depending on whether the N _U value and the N _L value can be obtained as described above.

Decoder according to the invention

Next, a decoder for decoding a context-based adaptive variable length coded (CAVLC) bitstream according to the present invention may include an input unit for receiving the bitstream, a decoding information storage unit, and an entropy decoder.

Specifically, the input unit of the decoder receives a bitstream for a first frame and a subsequent second frame composed of a plurality of blocks. The input unit of the decoder may include a frame buffer to store a predetermined number of frames.

Preferably, the decoder according to the present invention uses the encoding information of the block of the first frame at the same position as the target block of the second frame and the encoding information of the adjacent block of the already decoded second frame. And an entropy decoder that decodes the block.

The decoding information storage unit of the decoder stores decoding information generated while decoding the block of the first frame through the entropy decoder. The decoding information is already in front of (i) the number of nonzero coefficients in the block, (ii) the number of trailing ones in the block, (iii) the sign of the trailing one, and (iv) the nonzero coefficients in the block. Contains the total number of zeros. Preferably, the decoder according to the present invention uses the number of nonzero coefficients in a block among the decoding information.

As in the encoder, in the decoder, the adjacent block is preferably a block located on the left side and a block located on the upper side of the target block.

In addition, as already seen in the encoder, the reference value N for encoding the target block in the decoder may be represented by Equation 1 when both N _L and N _U values are obtained, and Equation 2 when N _U values cannot be obtained. If you can not get the N _L value can not be obtained when both the equation (3), the last value of N _L and N _U value N is set to N _P value.

Preferably, the entropy encoder of the encoder has the same number of variable coding tables as the entropy decoder of the decoder. For example, in the case of H.264, the intoder and the decoder may include the FLC having the six-bit xxxxyy form and the variable coding tables Num-VLC0, Num-VLC1, Num-VLC2, which are the four reference tables shown in Tables 4 to 6 described above. Include.

Preferably, the entropy decoder of the decoder uses Num-VLC0 when the N value is 0 or 1, uses Num-VLC1 when the N value is 2 or 3, and Num- when the N value is 4 to 7. VLC2 is used, and FLC is used when the N value is 8 or more.

Hereinafter, a method of decoding a context-based adaptive variable length coded bitstream will be described. As shown in FIG. 11, the method for decoding a context-based adaptive variable length coded bitstream according to the present invention includes the number of nonzero coefficients in an adjacent block of a target block represented by the bitstream and the target in a previous frame. Selecting (400) a lookup table to use from nonzero coefficients in the block at the same location as the block; Decoding (402) a number of nonzero coefficients and a number of trailing ones in the target block using the first portion of the bitstream and the lookup table; Decoding (404) the sign of trailing 1 from the bits of the bitstream; Decoding (406) non-zero coefficient magnitudes except trailing 1 from the bits of the bitstream; Decoding (408) the number of coefficients of zero located before the last non-zero coefficient; And decoding 410 the position of the zero coefficient that is located before the last non-zero coefficient.

Preferably, in the method for decoding a context-based adaptive variable length coded bitstream according to the present invention, the reference table includes four types of Num-VLC0, Num-VLC1, Num-VLC2, and FLC.

In addition, preferably in the method for decoding a context-based adaptive variable length coded bitstream according to the present invention, as described above, the reference table to be used is determined by using Equations 1 to 3 or by an N value where N = Np. do.

Video transmission system according to the present invention

The video transmission system according to the present invention uses the configuration of the encoder and decoder described above. Specifically, a video transmission system for transmitting and receiving video data according to the present invention includes an encoder for compressing video data through a series of routines and transmitting the video data in the form of a bitstream and another series of bitstreams received from the encoder. It includes a decoder to decode through to reconstruct the video data.

Furthermore, the encoder of the video transmission system according to the present invention encodes a block of a video frame by using the encoding information of the adjacent block of the block and the encoding information of the block at the same position of the previous frame.

In addition, the decoder of the video transmission system according to the present invention reconstructs a bitstream using encoding information of an adjacent block of the block and encoding information of a block located at the same position in a previous frame.

Of the present invention Example

Hereinafter, a preferred embodiment of the present invention will be described with reference to the drawings. This embodiment is only an aspect of the present invention and should not be understood as limiting the scope of the present invention.

In this embodiment, the technical idea of the present invention is applied to H.264. In the present embodiment, it is assumed that all blocks have a size of 4x4 pixels, and that encoding information of the n-1th frame is used for encoding the nth frame. In addition, when quantizing the target block, the quantization parameter QP is set to 20 (the size Qstep of the quantization step is 6.5). Therefore, using Table 7, the α value is 0.145 and the β value is 0.71.

First, a process of encoding a frame according to an embodiment of the present invention will be described. First, the encoder frequency-transforms the difference block of 4x4 pixel size of the n-1 th frame using the H matrix shown in FIG. The transformed matrix is quantized using a 6.5 quantization step. For the quantization block thus obtained, information about the number NumCoeff of non-zero coefficients of each block obtained in the context-based adaptive variable length encoding process is stored in a memory.

After encoding the n-1th frame and then encoding the nth frame, the process of encoding the n-1 frame is repeated as it is. A process of encoding in the n-th frame will be described in detail as follows.

When encoding block 506 of the n-1th frame, the upper block 502 and the left block 504 are already encoded, and N _U = 3, respectively, for the number of nonzero coefficients in the blocks obtained in the process. , N _L = 2, and assume that the number of nonzero coefficients of block 500 at the same location as block 506 in frame n-1 is Np = 0.

Therefore, when the reference value N of the block 506 is obtained by Equation 1, round (0.145 x 5 +0.71 x 0) = 1. As described in Table 1 above, the variable length coding table used for the value of N = 1 becomes Num-VLC0 as given in Table 4. Compared with the prior art, in the prior art, since N value is 3 in the formula N = round (N _L + N _U ) / 2, Num-VLC1 is used to encode the same block 506. Therefore, the embodiment of the present invention and the conventional H.264 use different variable length coding tables.

A matrix for the block 506 to be encoded in the nth frame is given as shown in Table 8 below.

0 3 -One 0 0 -One One 0 One 0 0 0 0 0 0 0

Scanning the blocks in Table 8 above in the order shown in Figure 3, 0, 3, 0, 1, -1, -1, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0 A bit string of is obtained. The nonzero coefficients in this matrix are 3, 1, -1, -1, and 1, so the number of nonzero coefficients (NumCoeff) is 5. In addition, trailing 1 has 1, -1, -1, 1, but since only three of them are considered, the number of trailing 1 (T1s) is three. Looking at Num-VLC0 given in Table 4 for the block of Table 8, the code value is 0001011, and this code value is the first code value of the bitstream.

Next, since the last trailing 1 value has a positive sign, a code value of 0 is assigned, and the other two trailing 1 have a negative value, so a code value of 1 is given. Accordingly, the bit string corresponding to the sign of trailing 1 is 011.

The nonzero coefficient before the three trailing ones is 1, so the code value is 1 using the Level_VLC0 table. The next nonzero coefficient is 3, so the code value is 0010 using the Level_VLC1 table. As described above, a method of generating a bit string for coefficient sizes using the Level_VLC table is described in detail in the H.264 standard, and thus a detailed description thereof will be omitted.

Next, we code for the number of zeros before the last trailing one. Since the bit stream before the last trailing 1 is 0 , 3, 0 , 1, -1, -1, 0 , there are a total of 3 coefficients. When this is encoded using a separate variable length encoding table, a value of 111 is designated. The method of generating a bit string for the number of coefficients that are zero is also described in detail in the H.264 standard, and thus a detailed description thereof will be omitted.

Next, if the position of 0 located before trailing 1 is encoded, 101101 is obtained. The method of generating a bit string for a zero coefficient before a nonzero coefficient is also described in detail in the H.264 standard, so a detailed description thereof will be omitted.

The bitstream obtained with respect to the block shown in Table 8 is subjected to this process, resulting in 0001011 011 10010 111 101101.

Hereinafter, an embodiment of receiving and decoding a bitstream transmitted from an encoder in the decoder will be described. In the decoder according to the embodiment of the present invention, the decoding has already been completed for the n-1th frame, and the decoding information for each block of the n-1th frame, that is, the number N non-zero coefficients (N _p ) is stored in the memory. Used to decode the nth frame.

When decoding the target block 506 of the n-th target frame, it is already decoded with respect to the block 500 of the n−1 th frame to obtain an N _p value. Similarly, the upper block 502 and the left block 504 of the target block 506 are also decoded to obtain the number N _U and N _L of non-zero coefficients. Thus, N _U when decrypting block 506. = 3, N _L = 2, N _P = 0, and the quantization parameter (QP) is passed to the entropy decoder through another bitstream. Therefore, the reference value N can be calculated from the equation (1). The calculated value is 1 as in the encoder.

From the reference value thus obtained, the variable length coding table to be used for decoding the block 506 is determined as Num-VLC0. Then, the number of nonzero coefficients (NumCoeff) and the number of trailing ones (T1s) in the first bit string 0001011 are obtained. Since the number of trailing ones is 3 in the above bit string, the next three bits refer to the sign of trailing one. Thus, the bit strings of 1, -1, -1 are reconstructed from 0001011 011.

The subsequent bit strings represent the magnitudes of the nonzero coefficients, so the nonzero bit sizes are 1 and 3. That is, the bit strings of 3, 1, -1, -1, 1 are reconstructed from 0001011 011 10010. The next 111 bits means that the number of zeros before the last one of the column is three, and 101101 tells us about the location of these zeros. As a result, the reconstructed bit string is 3, 0, 1, -1, -1, 0, 1, which is the same as the bit string encoded by the decoder. The remaining nine bits after the last one are filled with zeros.

Although in the above embodiment, the context-based adaptive variable length coding method according to the present invention is applied to the H.264 method, the present invention is not limited to the H.264 method, and a plurality of frames using the context-based adaptive variable length coding method are used. The present invention can be applied to various types of compression techniques used to transmit data consisting of data.

In addition, in the present embodiment, the variable length coding table specified in H.264 is used. However, even if the variable length coding table is composed of a number of variable length coding tables different from the H.264 standard or the same number of variable length coding tables is used, variable lengths having different code values are different. The encoding table can of course also be used.

The context-based adaptive variable length coding method according to the present invention can perform encoding more efficiently. The method according to the present invention can more accurately predict the number N of non-zero coefficients expected in the target block, thereby enabling more efficient context-based adaptive variable length encoding and decoding. Table 9 below shows the accuracy of the predicted number of nonzero coefficients when encoded by the conventional H.264 and the predicted number of nonzero coefficients when encoded according to the present invention. Comparisons are made with respect to the quantization parameter (QP).

As can be seen from Table 9, the context-based adaptive variable length coding method according to the present invention can be seen that the accuracy of predicting the reference value (N) is significantly superior to the conventional H.264.

According to this accuracy, the same video can be transmitted using a low bit transmission bandwidth. Table 10 below shows the bit transmission bandwidth required when the context-based adaptive variable length coding method according to the present invention is used and when the context-based adaptive variable length coding method according to the conventional H.264 is used.

As shown in Table 10 above, it can be seen that using the context-based adaptive variable length coding method according to the present invention, it is possible to transmit a video having the same content using a lower bit transmission bandwidth. Therefore, by using the context-based adaptive variable length coding method according to the present invention, a video can be transmitted at a higher compression rate.

Claims (44)

An encoder (30) for context-based adaptive variable length coding (CAVLC) of quantized transform coefficients of data comprising a first frame and a subsequent second frame, A previous frame encoding information storage unit for storing encoding information N _p obtained by context-based adaptive variable length encoding of the first frame; And The context-based adaptive variable length encoder 40 generates a reference value N for context-based adaptive variable length encoding of the second frame in consideration of the encoding information N _p stored in the previous frame encoding information storage unit. An encoder for context-based adaptive variable length encoding of the quantized transform coefficients of the data, characterized in that it comprises a. The method of claim 1, wherein the encoder A block divider for dividing each frame into a plurality of blocks; A frequency converter 34 for converting the data values of the divided blocks into a frequency domain; And And an quantization unit (36) for quantizing the data values of the block transformed into the frequency domain. 3. The method of claim 2, wherein the context-based adaptive variable length coding is performed in units of blocks of a frame, and when the target block 506 of the second frame is context-based adaptive variable length coding, the encoding information is frequency transformed. the number (N _P) of non-zero coefficients in the block 500 in the same position in the quantization block of a first frame with a position without considering the target block and the motion information of the second frame, the reference value (N) Is a number (N) of non-zero coefficients expected in the target block of the second frame, wherein the encoder is context-based adaptive variable length encoding of the quantized transform coefficients of the data. 4. The method of claim 3, wherein the number N of nonzero coefficients expected in the target block 506 is determined in block 504 of the second frame located to the left of the target block 506 of the second frame. The number of nonzero coefficients N _L and the number of nonzero coefficients N _U in the block 502 of the second frame located above the target block of the second frame are also calculated. An encoder for context-based adaptive variable length coding of quantized transform coefficients of the data. 5. The method of claim 4, wherein the number N of nonzero coefficients expected in the target block 506 is N = round {αx (N _U + N _L ) + βN _P } (where α and β are positive real numbers satisfying the relationship of 2α + β = 1). 6. The context-based adaptive variable of claim 5, wherein the values of α and β vary according to the Qstep of the quantization step used when the quantization unit 36 quantizes the data values of the block. Length Coding Encoder. The method of claim 6, When the magnitude (Qstep) of the quantization step is 0.625 to 4, the α value is 0.125 and the β value is 0.75, When the magnitude (Qstep) of the quantization step (Qstep) is 4.5 to 16, the α value is 0.145 and the β value is 0.71, And wherein the α value is 0.165 and the β value is 0.67 when the Q-step size of the quantization step is 18 to 224. 8. The encoder (30) according to any one of claims 3 to 7, wherein the encoder (30) stores a plurality of look-up tables, wherein the expected number of nonzero coefficients of the target block (506). Context-based adaptive variable length encoding encoder according to (N), wherein one of the plurality of lookup tables is selected to perform encoding of the target block (506). The context-based adaptive variable length encoding encoder according to claim 1, wherein the data is in H.264 format and the first frame and the second frame constitute a video in H.264 format. In the encoder 30 for converting data including a video signal consisting of a plurality of frames into a bitstream, An input frame storage unit 31 for storing data of a target frame input to the encoder 30; A reference frame storage unit 52 for storing data of a previous frame of the target frame; A motion estimation and compensator (54, 56) for selecting a block of the previous frame closest to the block of the target frame; A subtractor (32) for obtaining a block for a difference value between a block of a previous frame obtained by the motion estimation and compensation unit and a target block (506) of the target frame; A converter (34) for converting a block (Dn) of the difference value obtained by the subtractor into a frequency domain; A quantization unit 36 for quantizing a block transformed into the frequency domain by a predetermined quantum step size (Qstep); And A coding unit 40 for performing context-based adaptive variable length coding (CAVLC) on the quantized block in the quantization unit 36, In encoding the target block 506, the encoder 40 does not consider the encoding information N _U and N _L of the adjacent blocks 502 and 504 of the target block and the motion information in the previous frame. An encoder for converting data including a video signal consisting of a plurality of frames into a bitstream, characterized by using the encoding information (N _P ) of the block (500) at the same position as the target block. The encoder of claim 10, wherein the encoding information is a number NumCoeff of non-zero coefficients among coefficients included in a block. 12. The apparatus of claim 11, wherein the encoder 40 includes a plurality of variable length encoding reference tables, and the encoder 40 is in the previous frame with adjacent blocks 502 and 504 of the target block 506. One of the variable length encoding reference tables based on the reference value N of the target block obtained from the encoding information N _P of the block 500 at the same position as the target block 506 without considering motion information. And converting the data into a bitstream. 13. The encoder according to claim 12, wherein the adjacent block is a block (504) located above and a block (502) located above the target block. The method of claim 13, wherein the reference value N of the target block 506 is N = round {αx (N _U + N _L ) + βN _P } (where N _L is the number of nonzero coefficients in the block on the left in the same frame as the target block, and N _U is 0 in the block on the top in the same frame as the target block) N _P is the number of nonzero coefficients in the block at the same position as the target block in the previous frame, and α and β are positive real numbers that satisfy the relationship 2α + β = 1). An encoder for converting data into a bitstream, characterized in that determined by. 15. The context-based adaptive variable length encoding encoder of claim 14, wherein the values of α and β vary depending on the size of the quantization step (Qstep). The method of claim 15, When the magnitude (Qstep) of the quantization step is 0.625 to 4, the α value is 0.125 and the β value is 0.75, When the magnitude (Qstep) of the quantization step (Qstep) is 4.5 to 16, the α value is 0.145 and the β value is 0.71, And wherein the α value is 0.165 and the β value is 0.67 when the Q-step size of the quantization step is 18 to 224. In the decoder 100 for decoding a context-based adaptive variable length coded (CAVLC) bitstream, An input unit configured to receive a bitstream for a first frame and a subsequent second frame composed of a plurality of blocks; An encoding information storage unit which stores encoding information generated while decoding the block of the first frame; And The encoding information N _P of the block 500 of the first frame at the same position as the target block of the second frame and the encoding information N _U , N of the adjacent blocks 502, 504 of the second frame which have already been decoded. _L ) and an entropy decoder (110) for decoding the target block (506) of the second frame using the decoder of decoding a context-based adaptive variable length coded bitstream. 18. The decoder of claim 17 wherein the encoding information is the number of nonzero coefficients (NumCoeff) in the block when the block is decoded. 19. The decoder of claim 18 wherein the adjacent block is a block (504) located above and a block (502) located above the target block. The entropy decoder 110 of claim 19, wherein the entropy decoder 110 includes a plurality of variable coding tables, and the variable coding table for the target block of the second frame is N = round {αx (N _U). + N _L ) + βN _P } (where N _L is the number of nonzero coefficients in the block located to the left of the target block in the same frame as the target block, and N _U is the top of the target block in the same frame as the target block) Is the number of nonzero coefficients in the located block, N _P is the number of nonzero coefficients in the block at the same position as the target block in the previous frame, and α and β are positive quantities that satisfy the relationship 2α + β = 1 A decoder for decoding a context-based adaptive variable length coded bitstream. 21. The decoder of claim 20, wherein the entropy decoder comprises four variable coding tables. In the decoder for decoding a bitstream consisting of a plurality of slices, An input unit configured to receive a bitstream of a frame of an image including a plurality of blocks; And An entropy decoder 110 for reconstructing the transform block of the frame in units of blocks constituting the frame from the slice, When the entropy decoder 110 decodes the target block 506 of the frame, encoding information (N _U , N) of adjacent blocks 502, 504 of the target block 506 included in the slice and already decoded. _L ) and a plurality of slices, which are decoded by a look-up table selected by encoding information N _P of a block 500 at the same position as the target block in a previous frame. Decoder to decode the resulting bitstream. 23. The method of claim 22, wherein the lookup table is a variable length coded table and the coded information is a number of non-zero coefficients NumCoeff in the decoded blocks 500, 502, 504. Decoder to decode the composed bitstream. 24. The bitstream of claim 23, wherein the adjacent block is a block 504 located on the left side of the target block 506 and a block 502 located on the upper side of the target block. Decoder to decode. 25. The variable coding table for the target block 506 according to claim 24, wherein if the slice includes bitstreams for the left block 504 and the upper block 502 of the target block 506, N = round. {αx (N _U + N _L ) + βN _P } (where N _L is the number of nonzero coefficients in the block located to the left of the target block in the same frame as the target block, and N _U is the top of the target block in the same frame as the target block) Is the number of nonzero coefficients in the located block, N _P is the number of nonzero coefficients in the block at the same position as the target block in the previous frame, and α and β are positive quantities that satisfy the relationship 2α + β = 1 A decoder for decoding a bitstream consisting of a plurality of slices, characterized in that the real number (is a real number). 25. The target block 506 of claim 24, wherein if the slice contains a bitstream for the left block 504 of the target block but does not include the bitstream for the upper block 502 of the target block. Variable coding table for N = round (αN _L + βN _P ) (where N _L is the number of nonzero coefficients in the block located to the left of the target block within the same frame as the target block, and N _P is the previous Number of nonzero coefficients in a block at the same position as the target block in the frame, and α and β are positive real numbers satisfying the relationship of 2α + β = 1). Decoder to decode the composed bitstream. 25. The target block 506 of claim 24, wherein if the slice includes a bitstream for the upper block 502 of the target block but does not include the bitstream for the left block 504 of the target block. The variable coding table for N = round (αN _U + βN _P ), where N _U is the number of nonzero coefficients in the block located above the target block within the same frame as the target block, where N _P is the previous frame Is a number of nonzero coefficients in a block at the same position as the target block, and α and β are positive real numbers satisfying a relationship of 2α + β = 1). Decoder to decode the resulting bitstream. 25. The variable coding table for the target block 506 according to claim 24, wherein if the slice does not include bitstreams for the left block 504 and the upper block 502 of the target block, N = N _P ( Where N _P is the number of nonzero coefficients in the block at the same location as the target block in the previous frame). 29. The entropy decoder (110) according to any one of claims 25 to 28, wherein the entropy decoder (110) comprises four variable coding tables (Num-VLC0, Num-VLC1, Num-VLC2, FLC). A decoder for decoding a bitstream consisting of a plurality of slices. The method of claim 29, wherein the entropy decoder 110 uses Num-VLC0 when the N value is 0 or 1, uses Num-VLC1 when the N value is 2 or 3, and the N value is 4 Num-VLC2 is used in the range from 7 to 7, and FLC is used if the value of N is 8 or more. In the context-based adaptive variable length coding method of a plurality of frames of video data, Determining (282) the number of trailing ones (T1s) and the number of nonzero coefficients in the series of transform coefficients for the target block of the frame; Determining a variable length encoding table from the number of nonzero coefficients in the neighboring block of the target block and the number of nonzero coefficients in the block at the same position as the target block without considering motion information in a previous frame (284); And Obtaining a code value for the target block from the number of trailing ones and the number of nonzero coefficients in the selected variable length encoding table and the series of transform coefficients (286). A context-based adaptive variable length encoding method of the generated video data. The method of claim 31, wherein Incubating (254) the sign of trailing 1 in the series of transform coefficients; Encoding (256) a level value of the nonzero coefficients; Encoding (258) the number of all zero coefficients in front of the last non-zero coefficient in the series of transform coefficients; And And encoding (260) positions of all zero coefficients in front of the last non-zero coefficients. 33. The method of claim 32, wherein the adjacent blocks 502 and 504 are blocks that are encoded before the target block 506 in the order of encoding the blocks of the frame. . 34. The method of claim 33, wherein the adjacent block is a left block (504) and an upper block (502) of the target block. 35. The method of claim 34, wherein determining the variable length encoding table (284) comprises: Obtaining (302) a number N _U of nonzero coefficients in an upper block of the target block; Obtaining (304) the number N _L of nonzero coefficients in the left block of the target block; Obtaining the number N _P of non-zero coefficients in the block in the same location as the current block in the previous frame (306); Calculating a value of N = round {αx (N _U + N _L ) + βN _P }, wherein α and β are positive real numbers satisfying the relationship of 2α + β = 1; And And selecting (310) a variable length encoding table to be used based on the value of N. 35. The method of claim 34, wherein determining the variable length encoding table (284) comprises: Obtaining (304) the number N _L of nonzero coefficients in the left block of the target block; Obtaining the number N _P of non-zero coefficients in the block in the same location as the current block in the previous frame (306); N = round (αxN _L calculating (308) a value of + βxN _P (where α and β are positive real numbers satisfying a relationship of 2α + β = 1); And And selecting (310) a variable length encoding table to be used based on the value of N. 35. The method of claim 34, wherein determining the variable length encoding table (284) comprises: Obtaining (302) a number N _U of nonzero coefficients in an upper block of the target block; Obtaining the number N _P of non-zero coefficients in the block in the same location as the current block in the previous frame (306); N = round (αxN _U calculating a value of + βxN _P (where α and β are positive real numbers satisfying a relationship of 2α + β = 1) (308); And And selecting (310) a variable length encoding table to be used based on the value of N. 35. The method of claim 34, wherein determining the variable length encoding table (284) comprises: Obtaining the number N _P of non-zero coefficients in the block in the same location as the current block in the previous frame (306); And And selecting (310) a variable length encoding table to be used based on the value of N _P. 38. The method according to any one of claims 35 to 37, wherein the α and β values depend on the quantization step size (Qstep) used to obtain the transform coefficients of the block. How to. The method of claim 39, When the magnitude (Qstep) of the quantization step is 0.625 to 4, the α value is 0.125 and the β value is 0.75, When the magnitude (Qstep) of the quantization step (Qstep) is 4.5 to 16, the α value is 0.145 and the β value is 0.71, And the α value is 0.165 and the β value is 0.67 when the Q step size is 18 to 224. 4. A method for decoding a context-based adaptive variable length coded bitstream, In the adjacent blocks 502 and 504 of the target block 506 represented by the bitstream, the number of nonzero coefficients N _U , N _L and the same as the target block in position without considering the motion information in the previous frame. Selecting (400) a lookup table to use from non-zero coefficients at block 500 in position; Decoding (402) the number of nonzero coefficients (NumCoeff) and the number of trailing ones in the target block using the first portion of the bitstream and the lookup table; Decoding (404) the sign of trailing 1 from the bits of the bitstream; Decoding (406) non-zero coefficient magnitudes except trailing 1 from the bits of the bitstream; Decoding (408) the position of the zero coefficient that precedes the last non-zero coefficient; And Decoding (410) the position of the zero coefficient that is located before the last non-zero coefficient. 42. The method of claim 41, wherein the reference table is Num-VLC0, Num-VLC1, Num-VLC2, FLC four kinds, N = round {αx (N _U + N _L ) + βN _P } (N _L Is the number of nonzero coefficients in the block located to the left of the target block within the same frame as the target block, N _U is the number of nonzero coefficients in the block located above the target block within the same frame as the target block, N _P is the number of nonzero coefficients in the block at the same position as the target block in the previous frame, and α and β are positive real numbers that satisfy the relationship 2α + β = 1). Num-VLC0 if 0 or 1, Num-VLC1 if N is 2 or 3, Num-VLC3 if N is 4 to 7, and FLC reference table if N is 8 or more. A method for decoding a context-based adaptive variable length coded bitstream. 43. The method of claim 42, wherein the α and β values depend on the quantization step size (Qstep) for quantizing the block, wherein the α value is 0.125 and the β value is 0.625 to 4 when the Qstep size of the quantization step is 0.625 to 4. 0.75, the value of α is 0.145 when the quantization step (Qstep) is 4.5 to 16 and the value of β is 0.71. When the size of the quantization step (Qstep) is 18 to 224, the value of α is 0.165 and the value of β. Is 0.67. A method for decoding a context-based adaptive variable length coded bitstream. In the video transmission system for transmitting and receiving video data, Encoders 12 and 14 for compressing video data through a series of routines and transmitting them in the form of a bitstream; And Decoders 14 and 12 for reconstructing the video data by decoding the bitstream received from the encoder through another series of routines, The encoders 12 and 14 do not consider the encoding information N _U and N _L of adjacent blocks 502 and 504 of the block and the motion information in the previous frame when encoding a block 506 of a frame of the video. Encoding using the encoding information N _P of the block 500 at the same position The decoders 14 and 12 may encode encoding information N _U and N _L of the adjacent blocks 502 and 504 of the block 506 and encoding information N _P of the block 500 at the same position in the previous frame. The video transmission system for compressing and transmitting video data, characterized in that for reconstructing the bitstream using a.

Images