KR101325795B1 - Method and Apparatus for Decoding Macroblock Data in H.264/AVC - Google Patents

Abstract

Disclosed are a method and apparatus for decoding macroblock data in H.264 / AVC. In the present invention, in decoding a symbol from a bit stream compressed by the H.264 / AVC standard, an information extraction algorithm of neighboring blocks required for decoding of macroblock data is optimized. Accordingly, macroblock decoding can be made more simply and efficiently, and the amount of macroblock data to be stored in the buffer for decoding can be minimized.

H.264 / AVC, Macroblock, Neighbor Block, Picture Mode

Description

Decoding method and decoding apparatus for macroblock data in H.264 / ACC {Method and Apparatus for Decoding Macroblock Data in H.264 / AVC}

1 is a block diagram of a decoding apparatus according to an embodiment of the present invention;

2A and 2B are views provided to explain an operation of a neighbor information controller according to an embodiment of the present invention;

3 is a view provided to explain a method of selecting a neighboring block when the current block is not an edge block;

4 is a view provided to explain a method of selecting a neighboring block when the current block is an edge block and the current macroblock is not mbaff;

5 is a diagram provided to explain a method of selecting a neighboring block when a current macroblock is mbaff and a picture mode of a current macroblock and a neighboring macroblock is the same;

6 is a diagram provided to explain a method of selecting a neighboring block when a current macroblock is mbaff, a current macroblock is a frame mode, and a neighboring macroblock is a field mode;

7 is a diagram provided to explain a method of selecting a neighboring block when a current macroblock is mbaff, a current macroblock is a field mode, and a neighboring macroblock is a frame mode;

8 is a diagram provided to explain a method of obtaining a neighboring block B when the current macroblock is mbaff and the current macroblock is a Top macroblock of a macroblock pair;

9 is a diagram provided to explain a method of obtaining a neighboring block B when the current macroblock is mbaff and the current macroblock is a Bottom macroblock of a macroblock pair, and

FIG. 10 is a view provided to explain data stored in a buffer unit of FIG. 1.

The present invention provides a method of macroblock data in H.264 / AVC that efficiently manages and retrieves neighbor information necessary for decoding a macroblock in a bitstream among video decoder devices compressed to the H.264 / AVC standard. The present invention relates to a decoding method and a decoding apparatus.

H.264 / AVC uses neighbor information at many stages, such as selecting a Variable Length Coding (VLC) table and selecting a Context Adaptive Binary Arithmetic Coding (CABAC) model to reduce the number of bits representing a symbol in the bit stream. I use it.

H.264 / AVC standards include 16 × 16 Macro Blocks (MB), 8 × 8 luma (Luminance) blocks, 4 × 4 luma blocks, 4 × 4 chroma (Chrominance) blocks, and partitions ( Each section describes how to obtain the neighboring blocks or partitions of a partition. The following description is based on the fourth version of ISO / IEC 14496-10, which is a standard for H.264 / AVC.

The standard defines a difference of luma location (xD, yD) with neighboring blocks in the N direction with respect to the current macroblock as shown in Table 1 below. Here, the N direction includes directions A, B, C, and D. As shown in Table 1, A is the left direction of the current macroblock, B is the upper direction, C is the upper right direction, and D is the upper left direction. It defines a method of calculating the position (xN, yN) of the neighboring block from (xD, yD) according to whether the peripheral information mode to be obtained is a macroblock, a block, or a partition.

N xD yD A -One 0 B 0 -One C predPartWidth -One D -One -One

Once the position of the neighboring block (xN, yN) is determined, whether the current frame is a MacroBlock-Adaptive Frame / Field (MBAFF) or Picture Adaptive Frame / Field (PAFF), whether the current macroblock is a field structure, or a macroblock. Depending on whether the macroblock pair is a top macroblock, location information (xW, yW) is derived from the macroblock information mbAddrN and the macroblock, and the corresponding block or partition from (xW, yW). You can get index information of.

Such a conventional neighbor information detection method is complicated and performs calculations that are not necessary in various ways. It also does not mention the data in the current macroblock that must be stored in order to decode the symbols of the next macroblock.

SUMMARY OF THE INVENTION An object of the present invention is to decode macroblock data in H.264 / AVC, which efficiently manages and retrieves neighbor information required for decoding a macroblock in a bitstream compressed with the H.264 / AVC standard. A method and a decoding apparatus are provided.

In order to achieve the above object, a method of decoding macroblock data in H.264 / AVC according to the present invention comprises: receiving a bitstream compressed by the H.264 / AVC standard, by the received bitstream In the decoding process of each block of the current macroblock (MB: Macroblock) including the formed 4x4 block, at least one of the neighboring block A and the neighboring block B necessary for decoding the current block having the index (x, y). Extracting data from the pre-decoded and stored macroblock data, and parsing the current macroblock data by performing decoding on the current block from the received bitstream based on the extracted data. Steps.

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

1 is a block diagram of a decoding apparatus according to an embodiment of the present invention.

The decoding apparatus 100 of the present invention receives a bit stream, which is a video signal compressed by the H.264 / AVC standard, decodes and outputs data based on predetermined neighbor information. The decoding apparatus 100 of the present invention efficiently manages and retrieves neighbor information necessary for decoding a symbol in a bitstream compressed by the H.264 / AVC standard.

In the following description of the invention, parts not described otherwise may be supplemented by ISO / IEC 14496-10, which is a standard for H.264 / AVC.

H.264 / AVC defines four neighbor information, A, B, C, and D, depending on the direction, and the neighbor information according to the picture mode, macroblock (MB) mode, and block position to compress. Choose. A portion of the data of the macro block is then stored for decoding the symbols of the next macro block that comes later in time. The decoding apparatus 100 of the present invention can efficiently store only necessary data among data included in a macroblock and simply select neighbor information. Furthermore, the decoding apparatus 100 reduces the buffer size by storing data only for necessary blocks instead of storing data of all blocks in the macro block.

Referring to FIG. 1, the decoding apparatus 100 processes an MB data parser 110 that decodes macroblock data from an input bitstream, and processes neighboring information required by the MB data parser 110. The neighbor information controller 130 and a buffer unit 150 storing neighbor information are included.

The MB data parser 110 decodes current macroblock (currMB) data in the input bitstream compressed by the H.264 / AVC standard. If the MB data parser 110 needs information of a neighboring block (hereinafter, referred to as a 'neighbor block') in the decoding process of each block included in the current macroblock including the 4 × 4 block, the neighbor information controller 130 Request data of a neighboring block. The information on the neighboring block requested by the MB data parser unit 110 is compared with requesting information on a neighboring macroblock having a 4 × 4 block size in order to decode the data in the bitstream in H.264 / AVC.

The neighbor information controller 130 reads 'neighbor information', which is information about a neighboring block of the current block, from the buffer unit 150 based on the data type requested by the MB data parser 110 and the current block index. The MB data parser 110 transmits the data. The neighbor information controller 130 transmits only the information on the neighbor blocks A and B among the neighbor blocks A, B, C, and D provided by H.264 / AVC to the MB data parser 110. This will be described in detail below.

In addition, the neighbor information controller 130 extracts data necessary for decoding the macroblock data input after the MB data parser 110 decodes the current macroblock and stores the data in the buffer unit 150. .

The buffer unit 150 stores data of a neighbor block corresponding to the neighbor information. The buffer unit 150 provides the neighbor information to the neighbor information controller 130 according to a request of the neighbor information controller 130. When storing the neighbor information, the buffer unit 150 does not store all data of the neighboring macroblocks, but stores only the data of the lowest column and the rightmost row of one macroblock. The buffer unit 150 also stores a current picture mode, a macroblock mode, and the like, which are parameters required for selecting a neighboring block.

A method of selecting the neighbor information from the buffer unit 150 based on the data type requested by the MB data parser 110 and the block index will be described with reference to FIG. 2. . The operation of the buffer unit 150 will be described again below based on the description of FIG. 2.

2 is a view provided to explain the operation of the neighbor information control unit according to an embodiment of the present invention.

2 (a) relates to a method of obtaining a neighboring block in the A direction (left direction) of any block included in the current macroblock, and FIG. 2 (b) shows B of any block included in the current macroblock. It relates to a method for obtaining a neighboring block in an upward direction. In FIG. 2, only the neighboring blocks A and B are shown because the MB data parser 110 does not actually need the information about the neighboring block C or D in parsing the macroblock data from the bitstream. This is not presented in the standard. Therefore, the neighbor information controller 130 basically solves the problem of unnecessary data buffering and unnecessary calculation that may occur when parsing macroblock data according to a standard.

Hereinafter, the term 'current block' or 'neighborhood block' means a 4 × 4 luma block or a color signal (chroma) block, and the macroblock is a 16 × 16 (including 16 blocks) luminance signal or color signal. A block and macroblock pair represents a 16 × 32 (including two macroblock) luminance signal or color signal block. In the following description, "block" or "4x4 block" is used in the same sense.

Also, the neighbor data controller 130 decodes the block having the index (x, y) included in the current macroblock (hereinafter, referred to as 'current macroblock mb') in which the MB data parser 110 has a macroblock number of mb. In the case of requesting information on the neighboring block), a block having an index (x, y) is called a 'current block'. x and y are block indexes of the current block in a macroblock including 4x4 blocks.

mba is the number of the macroblock including the neighboring block A of the current block, and mbb is the number of the macroblock including the neighboring block B of the current block.

xa and ya are block indexes of the neighboring block A in the macroblock including the neighboring block A, and xb and yb are block indexes of the neighboring block B in the macroblock including the neighboring block B. Thus, x, y, xa, ya, xb, and yb correspond to integers from 0 to 3. All operations in FIG. 2 are integer operations. In this case, the integer operation means that only an integer is taken from the operation result of each step. For example, if y is 3 and mb is 5, y / 2 + (mb% 2) ㅧ 2 is 1 + 1 ㅧ 2 = 3. % Is the remainder divided by 5% 2 = 1.

Mode [] is a picture mode of the macroblock, and includes a field mode according to an interlaced method and a frame mode according to a progressive method. According to the mbaff compression method, Mode [mb] is the mode of the current macroblock, and mode [mb-2] is the macroblock number mb-2 of the macroblock, that is, the mode of the macroblock immediately to the left of the current macroblock. something to do. Mode [mb-2N] is a picture mode of a macroblock having the macroblock number mb-2N. Here, N is a width (PicWidthInMbs) of a picture calculated in units of macroblocks, and a macroblock column of one picture includes N macroblocks. In other words, N is the number divided by 16 divided by the width of the image.

Referring to FIG. 2, in determining a neighboring block A or B of a block included in a current macroblock, it is first determined whether the current block is an edge block of the current macroblock mb. Here, the edge block is a block belonging to the leftmost block row (first row) and the highest block column (first column) of the macroblock.

In determining the neighboring block, when the pixels of the picture are rearranged according to the field or frame mode of the current block with respect to the upper left point (or the upper left pixel) of the current block, The block including the pixel on the left side is determined as the neighboring block A, and the block including the pixel immediately above the neighboring block B. That is, if the current macroblock corresponds to the top field as the field mode, the neighboring pixel is selected from the top field image data of the current frame, and if the current macroblock is the bottom field, the neighboring pixel is selected from the bottom field image data. In the frame mode, when the current frame is considered to be a progressive mode, a block including adjacent pixels is a neighboring block. This method of determining applies to all cases of determining neighboring blocks in the present invention.

In case 1 of FIG. 2 (i.e., if it is not an edge block, if it is a non-boundary block), the block including the immediately left pixel of the upper left pixel of the current block is neighboring without having to consider anything else. The block including the pixel immediately above the upper left pixel of the current block is the neighboring block B. Therefore, the macroblock including the neighboring block A is the current macroblock mb, the index of the neighboring block A is (x-1, y), and the index of the neighboring block B is (x, y-1). Here, reference is made to FIG. 3 for description of case 1. FIG.

3 is a diagram provided to explain a method of selecting a neighboring block when the current block is not an edge block.

In FIG. 3, the smallest unit rectangle represents a 4x4 block, and the luminance signal luma component of the macroblock includes 16 blocks (4x4) in total. When the picture is in 4: 2: 0 color format, the chroma signal component for one macroblock is composed of four (2x2) blocks.

In the current macroblock, the second row 310, the third row 320 of the luma component, the rightmost block column 330, and the second 340 block row of the chroma signal are chroma. For the neighboring block A for the neighboring block belonging to the same macroblock is selected immediately. The arrow indicates the neighboring block of the current block, and the portion indicated by the thick line is the neighboring block. Similarly, neighboring blocks for the second column 350, the third column 360, the lowest block column 370 of the luminance signal luma, and the second column 380 of the color signal chroma in the current macroblock. B is the block immediately above belonging to the same macroblock is selected.

Referring again to FIG.

If the current block is an edge block or the current macroblock is not due to mbaff (case 2), the number mba of the macroblock to which neighboring block A belongs is mb-1, and the index of the neighboring block A (xa, ya) Becomes (3, y). The number mbb of the macroblock to which the neighboring block B belongs is mb-N, and the index (xb, yb) of the neighboring block B is (x, 3).

Further, even if the current block is an edge block and the current macroblock is caused by mbaff, when the neighboring block A is obtained when the current macroblock and the neighboring macroblock have the same picture mode (case 2), the macro to which the neighboring block A belongs is obtained. The number mba of the block becomes mb-2, and the index (xa, ya) of the neighboring block A becomes (3, y).

Here, reference is made to FIGS. 4 and 5 below for description of case 2. FIG.

FIG. 4 is a diagram provided to explain a method of selecting a neighboring block when the current block is an edge block and the current macroblock is not mbaff, and FIG. 5 is a current macroblock as a mbaff, and a current macroblock and a neighbor macro. It is a figure provided for description of the method of selecting a neighboring block when the picture mode of a block is the same.

If the current macroblock is not mbaff, as shown in FIG. 4, the neighboring macroblock is not mbaff. In (a) of FIG. 4, the neighboring block A for the blocks 410a and 410b of the leftmost row of the current macroblock is the blocks of the rightmost row of the left macroblock mb-1 of the current macroblock ( 420a and 420b, and the neighboring block B for the blocks 430a and 430b of the uppermost column of the current macroblock in FIG. 4B is the lowest column of the upper macroblock mb-N of the current macroblock. Blocks 440a and 440b.

FIG. 5 illustrates a case in which the current macroblock is mbaff and is in field mode, and the picture mode of the neighboring macroblock is also in field mode. Block A becomes the blocks of the rightmost row of the left macroblocks 530a and 540a of each current macroblock.

Referring to FIG. 2 again, a method of obtaining a neighboring block A when the current macroblock is due to mbaff and the picture modes of the current macroblock and the neighboring macroblock are different from each other will be described.

First, referring to (a) of FIG. 2, the neighboring block A is obtained when the current macroblock is in frame mode and the neighboring macroblock is in field mode (case 3). In other words, if Mode [mb] = frame and Mode [mb-2] = field.

In case 3, the macroblock mba to which neighboring block A belongs is mb-2- (mb% 2). The index (xa, ya) of the neighboring block A is shown in Equation 1 below.

An example of case 3 will be described with reference to FIG. 6.

FIG. 6 is a diagram provided to explain a method of selecting a neighboring block when a current macroblock is mbaff, a current macroble is a frame mode, and a neighboring macroblock is a field mode.

Referring to FIG. 6, the current macroblock pairs 610a and 620a include a top macroblock 610a and a bottom macroblock 620a. Since the current macroblock pairs 610a and 620a are in the frame mode, the upper left point of each block of the leftmost row of the current top macroblock 610a and the current bottom macroblock 620a becomes the pixel of the Top field. Since the neighboring macroblock is the field mode, the neighboring top macroblock 630a of the neighboring macroblock pairs 630a and 640a includes the top field pixel. Therefore, the neighbor block A of each of the blocks in the leftmost row of the current top macroblock 610a and the current bottom macroblock 620a is selected from the top macrofield 630a corresponding to the Top field among the neighboring macroblock pairs 630a and 640a. Is saved. If this is expressed as an equation, it is the same as Equation 1 shown in FIG.

Referring to (a) of FIG. 2 again, the neighboring block A is obtained when the current macroblock is the field mode and the neighboring macroblock is the frame mode (case 4). In other words, Mode [mb] = field and Mode [mb-2] = frame.

In Case 4, the macroblock mba to which the neighboring block A belongs is mb-2- (mb% 2) + (y / 2). And the index (xa, ya) of the neighboring block A is shown in the following equation (2).

,

if

,

else,

An example of case 4 will be described with reference to FIG. 7.

FIG. 7 is a diagram provided to explain a method of selecting a neighboring block when a current macroblock is mbaff, a current macroble is a field mode, and a neighboring macroblock is a frame mode.

Referring to FIG. 7, the current macroblock pairs 710a and 720a include a top macroblock 710a and a bottom macroblock 720a. Since the current macroblock pairs 710a and 720a are in the field mode, the upper left points of the blocks in the leftmost row of the current Top macroblock 710a are all pixels of the Top field. On the contrary, the upper left point of each block of the leftmost row of the current bottom macroblock 720a becomes the pixel of the Bottom field. In addition, since the neighboring macroblock pair is a frame mode, the macroblocks 730a and 740a of the neighboring macroblock pair include both a top field pixel and a bottom field pixel, respectively. Therefore, the neighbor block A of each block of the leftmost row of the current top macroblock 710a is obtained from a macrofield including corresponding Top field pixels among the neighboring macroblock pairs 730a and 740a.

In addition, the neighbor block A of each block in the leftmost row of the current bottom macroblock 720a is obtained from a macrofield including corresponding bottom field pixels among the neighboring macroblock pairs 730a and 740a. 7 shows these matters, and this is expressed by Equation 2 below.

This time, a method of obtaining a neighboring block B in cases other than cases 1 and 2 will be described with reference to FIG. The method of obtaining the neighbor block B largely depends on whether the current macroblock is a top macroblock or a bottom macroblock of a macroblock pair.

First, a case in which the current macroblock for which the neighboring block B is obtained is the top macroblock of the macroblock pair (case 5) will be described. Case 5 is further subdivided into cases in which both the current macroblock pair and the upper macroblock pair are in field mode (case 5-1) and otherwise (case 5-2). In other words, Mode [mb] = field, Mode [mb-2N] = field, and when it is not.

In case 5-1, the macroblock mbb to which the neighboring block B belongs is mb-2N. In case 5-2, the macroblock mbb to which the neighboring block B belongs is mb-2N + 1. The indexes (xb, yb) of the neighboring block B are all (x, 3).

An example of Case 5 is described with reference to FIG.

8 is a diagram provided to explain a method of obtaining a neighboring block B when the current macroblock is mbaff and the current macroblock is a top macroblock of a macroblock pair.

FIG. 8A illustrates a case in which both the current macroblock pair and the upper macroblock pair are in the field mode (case 5-1), and FIG. 8B illustrates both the current macroblock pair and the upper macroblock pair in frame mode. (Case 5-2).

Referring to FIG. 8A, since the current macroblock pair is in field mode, the upper left point of each block of the uppermost column of the top macroblocks 830a and 830b for which the neighbor block B is to be obtained corresponds to the top field pixel. Since the upper macroblock pairs 810 and 820 are also in the field mode, the upper top macroblocks 810a and 810b correspond to the top field pixels. Therefore, since the neighboring block B of each block of the top row of the top macroblocks 830a and 830b should be obtained from the upper top macroblocks 810a and 810b, the macroblock mbb to which the neighboring block B belongs becomes mb-2N and the neighboring block. The index (xb, yb) of B becomes (x, 3).

Referring to FIG. 8B, since the current macroblock pair is in frame mode, the upper left point of each block of the top column of the top macroblocks 860a and 860b for which the neighbor block B is to be obtained corresponds to the top field pixel. Since the upper macroblock pairs 840 and 850 are also in the field mode, corresponding to the top field pixels are present in both the upper top macroblocks 840a and 840b and the upper bottom macroblocks 850a and 850b. Therefore, the immediate upper block, which is the neighboring block B of each block of the top row of the top macroblocks 860a and 860b, must be obtained from the upper bottom macroblocks 850a and 850b, so that the macroblock mbb to which the neighboring block B belongs is mb-2N +. 1, and the neighboring index B (xb, yb) becomes (x, 3).

Finally, in the method of obtaining the neighboring block B, the case in which the current macroblock is the bottom macroblock of the current macroblock pair (case 6) will be described based on FIG.

Case 6 is a case in which the current macroblock has an mbaff syntax structure and is a Bottom macroblock of the current macroblock pair. If the current macroblock for which the neighbor block B is to be obtained is a bottom macroblock, the picture mode of the current macroblock is important. That is, the neighboring upper macroblock mbb is different according to the case where the picture mode of the current macroblock is the frame mode (case 6-1) and the field mode (case 6-2).

When the current macroblock is in frame mode as in case 6-1, the macroblock including the neighboring block B becomes the top macroblock of the current macroblock pair.

As in case 6-2, if the current macroblock is in field mode, the current macroblock corresponds to the bottom field pixel. Therefore, neighboring block B cannot be obtained from the top macroblock of the current macroblock pair as in case 6-1. It must be obtained from the bottom macroblock of the block pair.

Therefore, in case 6-1, the macroblock mbb to which the neighboring block B belongs is mb-1. In case 6-2, the macroblock mbb to which the neighboring block B belongs is mb-2N. The indexes (xb, yb) of the neighboring block B are all (x, 3).

An example of case 6 will be described with reference to FIG.

FIG. 9 is a diagram provided to explain a method of obtaining a neighboring block B when the current macroblock is mbaff and the current macroblock is a bottom macroblock of a macroblock pair.

FIG. 9A illustrates a case in which the current macroblock is in frame mode (case 6-1), and FIG. 9B illustrates a case in which the current macroblock is in field mode (case 6-2).

Referring to FIG. 9A, since the current macroblock is a frame mode, the upper left point of each block of the uppermost column of the bottom macroblocks 930a and 930b for which the neighboring block B is to be obtained corresponds to the top field pixel. Each block in the lowermost column of the top macroblocks 920a and 920b of the current macroblock pair becomes the neighboring block B to be obtained.

Referring to FIG. 9B, since the current macroblock is a field mode, the upper left point of each block of the uppermost column of the bottom macroblocks 960a and 960b for which the neighboring block B is to be obtained corresponds to the bottom field pixel. Since the top macroblocks 950a and 950b of the current macroblock pair correspond to the top field pixels, the neighbor block B cannot be obtained here. Therefore, since the neighboring block B must be obtained from the upper bottom macroblocks 930a and 930b, as shown in FIG. 9B, the macroblock mbb to which the neighboring block B belongs becomes mb-2N, and the indexes of the neighboring block B are indexes (xb, yb). ) Becomes (x, 3).

In the above, the method for obtaining neighboring blocks A and B according to the present invention has been described with reference to FIG. 2, and the operation of the neighbor information controller 130 and the buffer unit 150 will be described again based on the method.

The neighbor information controller 130, using FIG. 2, has a macroblock number and a picture mode of the current macroblock, a block number for which a neighbor block is to be obtained, and a neighboring macroblock mode. The numbers are easy to find.

3 to 9, it can be seen that blocks selected as neighboring blocks for the edge block of the current macroblock are blocks included in the rightmost row and the bottommost column of the macroblock, respectively.

Therefore, the buffer unit 150 stores only data corresponding to the rightmost row and the bottommost column of each macroblock among the outputs of the MB data parser 110. In addition, since neighboring blocks A and B belong to a maximum second macroblock, the buffer unit 150 stores data for up to 2N macroblocks. That is, the size of the efficient buffer unit 150 is 2N, and the data of the macroblock to be stored in the buffer unit 150 stores only data of blocks included in the rightmost row and the bottommost column. 10 illustrates data stored in the buffer unit 150.

FIG. 10 is a view provided to explain data stored in a buffer unit of FIG. 1.

The matrixes of FIGS. 10A to 10C are one picture and include N macroblocks in the horizontal direction.

(A) of FIG. 10 illustrates a case where the mbaff syntax value is '0' (that is, not mbaff), and (b) indicates that the mbaff syntax value is '1', and the macroblock currently being decoded is the Top macroblock. And (c) indicates that when the mbaff syntax value is '1' and the macroblock currently being decoded is a bottom macroblock, the entire data stored in the buffer unit 150 is indicated by a thick line.

Referring to FIG. 10, when the decoding of the current macroblock is completed, the buffer unit 150 may maximize the macroblock for decoding of the macroblock having the macroblock number mb + 1 (the macroblock to be decoded in the next step). The data corresponding to the macroblocks mb-2N + 1 to mb can be buffered, and the macroblock data before the macroblock number mb-2N + 1 can be deleted. Accordingly, the buffer unit 150 may buffer data necessary for the MB data parser 110 to decode using less memory.

The invention can be implemented in methods, devices and systems. In addition, when the present invention is implemented in computer software, the components of the present invention may be replaced with code segments necessary for performing necessary operations. The program or code segment may be stored in a medium that can be processed by a microprocessor and transmitted as computer data coupled with carrier waves via a transmission medium or communication network.

The media that can be processed by the microprocessor include electronic circuits, semiconductor memory devices, ROMs, flash memory, electrically erasable programmable read-only memory (EEPROM), floppy disks, optical disks, and hard disks. (Hard) Includes the ability to transmit and store information such as disks, fiber optics, wireless networks, and the like. Computer data also includes data that can be transmitted over electrical network channels, optical fibers, electromagnetic fields, wireless networks, and the like.

In addition, although the preferred embodiment of the present invention has been shown and described above, the present invention is not limited to the above-described specific embodiment, the technical field to which the invention belongs without departing from the spirit of the invention claimed in the claims. Of course, various modifications can be made by those skilled in the art, and these modifications should not be individually understood from the technical spirit or the prospect of the present invention.

As described in detail above, according to the method of storing and retrieving neighboring block information according to the present invention, it is possible to efficiently manage neighboring information necessary for decoding a macroblock in a bitstream compressed by the H.264 / AVC standard. You can search.

The method of the present invention can efficiently manage the buffers by buffering only the data necessary for parsing the macroblock data from the bitstream. In other words, the present invention reveals that neighboring blocks A, B, C, and D are not all necessary for parsing macroblock data, and buffer and manage macroblock data using only neighboring blocks A and B.

Based on the simple search algorithm proposed by the present invention, more efficient and improved decoding performance can be expected.

Claims (22)

Receiving a bit stream compressed to the H.264 / AVC standard; In the decoding process of each block of the current macroblock (MB: Macroblock) including the 4 × 4 block formed by the received bitstream, the neighboring block A (necessary for decoding the current block having an index (x, y)) ( Extracting data of at least one of a neighboring block in a left direction of the current block) and a neighboring block B (neighboring block in an upper direction of the current block) from previously decoded and stored macroblock data; And Parsing the current macroblock data by decoding the current block from the received bitstream based on the extracted data. Decoding method. The method of claim 1, Wherein the extracting comprises: The current block is included in the leftmost row of the current macroblock according to MBAFF (MacroBlock-Adaptive Frame / Field) mode, the current macroblock whose number of macroblocks is mb is frame mode data, and the macroblock If the macroblock whose number is mb-2 is field mode data, Extracting the block having the index (xa, ya) included in the macroblock with the macroblock number mba as the neighboring block A, Mba, xa, and ya are each represented by an integer operation. mba = mb-2- (mb% 2), xa = 3, A method of decoding macroblock data in H.264 / AVC, wherein ya = y / 2 + (mb% 2) × 2. The method of claim 1, Wherein the extracting comprises: The current block is included in the leftmost row of the current macroblock according to the MBAFF mode, the current macroblock whose number of macroblocks is mb is field mode data, and the macroblock whose number of macroblocks is mb-2 is a frame. For mode data, Extracting the block having the index (xa, ya) included in the macroblock with the macroblock number mba as the neighboring block A, Mba, xa, and ya are each represented by an integer operation. mba = mb-2- (mb% 2) + y / 2, xa = 3, A method of decoding macroblock data in H.264 / AVC, wherein ya = y × 2 if y <2, else ya = (y-2) × 2. The method of claim 1, Wherein the extracting comprises: The current block is included in the leftmost row of the current macroblock according to the MBAFF mode, and the current macroblock in which the number of the macroblock is mb and the macroblock in which the macroblock number is mb-2 are the same in the field mode or the frame. For mode data, Extracting the block having the index (xa, ya) included in the macroblock with the macroblock number mba as the neighboring block A, Mba, xa, and ya are each represented by an integer operation. mba = mb-2, xa = 3, A method of decoding macroblock data in H.264 / AVC, wherein ya = y. The method of claim 1, Wherein the extracting comprises: If the current block is included in the leftmost row of the current macroblock, and the current macroblock is not due to the MBAFF mode, Extracting the block having the index (xa, ya) included in the macroblock with the macroblock number mba as the neighboring block A, Mba, xa, and ya are each represented by an integer operation. mba = mb-1, xa = 3, A method of decoding macroblock data in H.264 / AVC, wherein ya = y. The method of claim 1, Wherein the extracting comprises: The current macroblock is a top macroblock according to the MBAFF mode, the current block is included in the top row of the current macroblock, and the number of the current macroblock and the macroblock in which the number of the macroblock is mb is mb-2N. If the macroblock is the same as the field mode data, where N is the width of the picture calculated in units of macroblocks, Extracting the block having the index (xb, yb) included in the macroblock with the macroblock number mbb as the neighboring block B, The mbb, xb, and yb are each an integer operation mbb = mb-2N, xb = x, A method of decoding macroblock data in H.264 / AVC, wherein yb = 3. The method of claim 1, Wherein the extracting comprises: The current macroblock is a top macroblock according to the MBAFF mode, the current block is included in the top row of the current macroblock, and the number of the current macroblock and the macroblock in which the number of the macroblock is mb is mb-2N. If at least one of the macroblocks is frame mode data, where N is the width of the picture calculated in units of macroblocks, Extracting the block having the index (xb, yb) included in the macroblock with the macroblock number mbb as the neighboring block B, The mbb, xb, and yb are each an integer operation mbb = mb-2N + 1, xb = x, A method of decoding macroblock data in H.264 / AVC, wherein yb = 3. The method of claim 1, Wherein the extracting comprises: When the current macroblock is a Bottom macroblock according to the MBAFF mode, the current block is included in the top column of the current macroblock, and the current macroblock whose number of macroblocks is mb is frame mode data. Extracting the block having the index (xb, yb) included in the macroblock with the macroblock number mbb as the neighboring block B, The mbb, xb, and yb are each an integer operation mbb = mb-1, xb = x, A method of decoding macroblock data in H.264 / AVC, wherein yb = 3. The method of claim 1, Wherein the extracting comprises: When the current macroblock is a Bottom macroblock according to MBAFF mode, the current block is included in the top column of the current macroblock, and the current macroblock whose number of macroblocks is mb is field mode data. Extracting the block having the index (xb, yb) included in the macroblock with the macroblock number mbb as the neighboring block B, The mbb, xb, and yb are each an integer operation mbb = mb-2N (where N is the width of the picture calculated in macroblocks), xb = x, A method of decoding macroblock data in H.264 / AVC, wherein yb = 3. The method of claim 1, Wherein the extracting comprises: If the current block is included in the top column of the current macroblock, and the current macroblock is not due to the MBAFF mode, Extracting the block having the index (xb, yb) included in the macroblock with the macroblock number mbb as the neighboring block B, The mbb, xb, and yb are each an integer operation mbb = mb-N (where N is the width of the picture calculated in macroblocks), xb = x, A method of decoding macroblock data in H.264 / AVC, wherein yb = 3. The method of claim 1, By storing the current macroblock data decoded in the parsing step, buffering data corresponding to macroblocks from macroblock numbers mb-2N + 1 to mb for decoding a macroblock with macroblock number mb + 1. And decoding the macroblock data in H.264 / AVC. Receive a bitstream compressed in the H.264 / AVC standard, decode the received bitstream to parse 4 × 4 blocks of macroblock data, and index (x, y) in the current macroblock that is the macroblock to be parsed. Based on data of at least one of neighbor block A (neighboring block in the left direction of the current block) and neighbor block B (neighboring block in the upper direction of the current block) necessary for decoding of the current block. An MB data parser configured to parse the current macroblock data by decoding a current block; A buffer unit for storing data on at least one of the pre-decoded and parsed macroblocks; And A neighbor information control unit for extracting data for at least one of the neighboring block A and the neighboring block B from the macroblock data stored in the buffer unit and transferring the extracted data to the MB data parser according to the request of the MB data parser; Macroblock data decoding apparatus in H.264 / AVC made. 13. The method of claim 12, The neighbor information control unit, The current block is included in the leftmost row of the current macroblock according to MBAFF (MacroBlock-Adaptive Frame / Field) mode, the current macroblock whose number of macroblocks is mb is frame mode data, and the macroblock If the macroblock whose number is mb-2 is field mode data, A block having an index (xa, ya) included in a macroblock having a macroblock number mba is designated as the neighboring block A and extracted from the buffer unit. Mba, xa, and ya are each represented by an integer operation. mba = mb-2- (mb% 2), xa = 3, A device for decoding macroblock data in H.264 / AVC, wherein ya = (y / 2) + (mb% 2) × 2. The method of claim 12, The neighbor information control unit, The current block is included in the leftmost row of the current macroblock according to the MBAFF mode, the current macroblock whose number of macroblocks is mb is field mode data, and the macroblock whose number of macroblocks is mb-2 is a frame. For mode data, The block having the index (xa, ya) included in the macroblock with the macroblock number mba is designated as the neighboring block A and extracted from the buffer unit. Mba, xa, and ya are each represented by an integer operation. mba = mb-2- (mb% 2) + y / 2, xa = 3, A device for decoding macroblock data in H.264 / AVC, wherein ya = y × 2 if y <2 and else ya = (y-2) × 2. 13. The method of claim 12, The neighbor information control unit, The current block is included in the leftmost row of the current macroblock according to the MBAFF mode, and the current macroblock in which the number of the macroblock is mb and the macroblock in which the macroblock number is mb-2 are the same in the field mode or the frame. For mode data, Extracting the block having the index (xa, ya) included in the macroblock with the macroblock number mba from the buffer unit as the neighboring block A, Mba, xa, and ya are each represented by an integer operation. mba = mb-2, xa = 3, A device for decoding macroblock data in H.264 / AVC, wherein ya = y. 13. The method of claim 12, The neighbor information control unit, If the current block is included in the leftmost row of the current macroblock, and the current macroblock is not due to the MBAFF mode, Extracting the block having the index (xa, ya) included in the macroblock with the macroblock number mba from the buffer unit as the neighboring block A, Mba, xa, and ya are each represented by an integer operation. mba = mb-1, xa = 3, A device for decoding macroblock data in H.264 / AVC, wherein ya = y. The method of claim 12, The neighbor information control unit, The current macroblock is a top macroblock according to the MBAFF mode, the current block is included in the top row of the current macroblock, and the number of the current macroblock and the macroblock in which the number of the macroblock is mb is mb-2N. If the macroblock is the same as the field mode data, where N is the width of the picture calculated in units of macroblocks, Extracting a block of index (xb, yb) included in a macroblock of macroblock number mbb from the buffer unit as the neighboring block B, The mbb, xb, and yb are each an integer operation mbb = mb-2N, xb = x, and yb = 3, the apparatus for decoding macroblock data in H.264 / AVC. 13. The method of claim 12, The neighbor information control unit, The current macroblock is a top macroblock according to the MBAFF mode, the current block is included in the top row of the current macroblock, and the number of the current macroblock and the macroblock in which the number of the macroblock is mb is mb-2N. If at least one of the macroblocks is frame mode data, where N is the width of the picture calculated in units of macroblocks, Extracting a block of index (xb, yb) included in a macroblock of macroblock number mbb from the buffer unit as the neighboring block B, The mbb, xb, and yb are each an integer operation mbb = mb-2N + 1, xb = x, and yb = 3, the apparatus for decoding macroblock data in H.264 / AVC. 13. The method of claim 12, The neighbor information control unit, When the current macroblock is a Bottom macroblock according to the MBAFF mode, the current block is included in the top column of the current macroblock, and the current macroblock whose number of macroblocks is mb is frame mode data. A block having an index (xb, yb) included in a macroblock with the macroblock number mbb is extracted as the neighboring block B, and the mbb, xb, and yb are each formed by integer operations. mbb = mb-1, xb = x, and yb = 3, the apparatus for decoding macroblock data in H.264 / AVC. 13. The method of claim 12, The neighbor information control unit, When the current macroblock is a Bottom macroblock according to MBAFF mode, the current block is included in the top column of the current macroblock, and the current macroblock whose number of macroblocks is mb is field mode data. Extracting a block of index (xb, yb) included in a macroblock of macroblock number mbb from the buffer unit as the neighboring block B, The mbb, xb, and yb are each an integer operation mbb = mb-2N (where N is the width of the picture calculated in macroblocks), xb = x, and yb = 3, the apparatus for decoding macroblock data in H.264 / AVC. 13. The method of claim 12, The neighbor information control unit, If the current block is included in the top column of the current macroblock, and the current macroblock is not due to the MBAFF mode, Extracting a block of index (xb, yb) included in a macroblock of macroblock number mbb from the buffer unit as the neighboring block B, The mbb, xb, and yb are each an integer operation mbb = mb-N (where N is the width of the picture calculated in macroblocks), xb = x, and yb = 3, the apparatus for decoding macroblock data in H.264 / AVC. 13. The method of claim 12, When the MB data parser completes decoding the current macroblock data, the neighbor information controller receives the decoded current macroblock data and delivers the decoded current macroblock data to the buffer unit. The buffer unit stores the decoded current macroblock data to buffer data corresponding to macroblocks from macroblock numbers mb-2N + 1 to mb for decoding a macroblock having macroblock number mb + 1. An apparatus for decoding macroblock data in H.264 / AVC.

Images