JPH0879767A - Moving vector coding and decoding system - Google Patents

Abstract

PURPOSE: To keep compatibility by reducing information quantity of moving image data even when a system is integrated into a syntax of a conventional MPEG system. CONSTITUTION: Digital image data are stored in a current frame memory 1 and fed to a preceding frame memory 6 and a macro block processing unit 2 dividing image data into plural macro blocks. A comparison macro block generator 7 generates a comparison macro block based on a retrieval point and preceding frame image data. A moving vector generator 3 generates a moving vector of a current macro block based on macro block data and the comparison macro block. In order to make coding less to the moving vector, a difference of adjacent macro blocks in horizontal and vertical directions with respect to the moving vector is taken and when the difference is small, the two macro blocks are regarded to represent the same object and grouped resulting that a moving vector is coded only once in one group.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a motion vector coding and decoding system, and more particularly to a motion vector coding and decoding system for coding and decoding a motion vector of a digital moving image.

[0002]

2. Description of the Related Art ISO is ISO / IEC 13818-2, MPE as a method for encoding digital moving images.
G2 (Moving Picture Expert Group 2) is being standardized. The MPEG2 coding method is basically based on dividing a frame into rectangular blocks (macroblocks) and performing DCT (discrete cosine transform) in units of blocks obtained by dividing the macroblock into four blocks. Frame encoding methods can be divided into intraframe encoding in which one frame is independently encoded and interframe predictive encoding in which a previous frame is used as a prediction frame. In a frame in which interframe coding is performed (hereinafter referred to as interframe), a motion from a reference frame is detected in macroblock units, coded as a motion vector, and transmitted together with pixel information.

Here, many methods for obtaining a motion vector have been reported, and the most general method is the full search method.
In this full search method, the similarity between the block of the previous frame and the block of the current frame is measured at all positions within the range to be searched (block matching), and the position with the highest similarity is set as the motion vector. To do. Note that the method of obtaining the motion vector is outside the MPEG standard. MP
If the standardization of the EG system advances, moving image data can be transmitted / received in any media without worrying about the system. It is also a great method for compressing moving images with a high compression rate.

By the way, as a motion vector coding technique, "a motion vector detecting method of an object on a screen" (hereinafter referred to as a first method) described in Japanese Patent Laid-Open No. 61-200789 and Japanese Patent Laid-Open No. 134085/1990. A second method "motion compensation coding method" (hereinafter referred to as the second method) described in Japanese Patent Laid-Open Publication has been conventionally proposed. Here, the first method is intended to reduce error detection by associating the motion vector with the vectors of the surrounding blocks, and is determined in advance by using the motion vectors of the blocks around the input block. The motion vector obtained by the motion vector detection unit is corrected according to the conditions. The second method encodes the differential motion vector with the adjacent block when the detected motion vector compensation error is smaller than the compensation error due to the motion vector of the adjacent block, and in other cases, the adjacent motion vector is not encoded. The motion vector of the block is used as the motion vector. That is, when the motion vector can be regarded as equivalent to the motion vector of the adjacent block, the motion vector is not encoded.

[0005]

If the standardization of the MPEG system is advanced, moving image data can be transmitted / received in any media without worrying about the system. It is also a great method for compressing moving images with a high compression rate. However, it is very difficult to transmit a color moving image in real time with a current general wired / wireless transmission circuit, and the code amount becomes enormous even if the MPEG system is used. On the other hand, the first method conventionally proposed for motion vector coding can reduce error detection by associating a motion vector with a vector of a surrounding block, but it is possible to reduce the code amount. could not. In the second method, when the motion vector can be regarded as equivalent to the motion vector of the adjacent block, the code amount of the motion vector is not coded so that the code amount can be reduced. It is necessary to calculate the motion vector predicted from the motion vector of the adjacent block every time, and the amount of calculation is considered to be enormous.

Therefore, an object of the present invention is to provide a motion vector coding and decoding system capable of reducing the information amount of moving image data. Also, conventional MP
It is an object of the present invention to provide a motion vector coding and decoding system that can maintain compatibility even when incorporated in the EG system syntax. Furthermore, it is another object of the present invention to provide a motion vector coding method capable of reducing the code amount while suppressing an increase in the calculation amount.

[0007]

According to a first aspect of the present invention, an input digital video signal is divided into a plurality of blocks, the blocks are sequentially scanned, and each block is subjected to block matching with a previous frame. In the method of detecting the motion vector and transmitting the obtained motion vector together with the pixel information of the image, the motion vector obtained in the current block and the motion vector in the adjacent block are compared, and both the motion vectors can be regarded as equal. A comparison determination means for determining whether the motion vector is a motion vector and a coding means for coding the direction of the adjacent block are not encoded when the comparison determination means determines that the motion vectors are equivalent. Prepare for the encoding method. According to a second aspect of the present invention, in the method according to the first aspect, the comparison / determination means sets the block in the direction in which the vertical scanning is started as the first block and starts the horizontal scanning when comparing with the adjacent block. The block in the above direction is the second block, and the block in the direction in which the horizontal scanning is finished is the third block. According to a third aspect of the present invention, in the method according to the first or second aspect, when the comparison / determination means makes a comparison / determination with an adjacent block in a direction in which scanning in the horizontal direction is terminated, the motion vector of the adjacent block is determined. After making a decision, make a comparative judgment. According to the fourth aspect of the present invention, the coded receiving means for receiving the motion vector and the direction of the adjacent block, and the block in the referring direction are sequentially traced from the block of the direction received by this receiving means to obtain the motion vector. When a block in which is present is referred to, a motion vector decoding method is provided with a decoding means for decoding the motion vector of the block as the motion vector of the current block.

[0008]

In the MPEG system, a motion vector is obtained for each macroblock in an inter frame and the difference between the motion vector and the motion vector of an adjacent macroblock is coded, whereas in the present system, this motion vector is coded as much as possible. In order to prevent this, the difference between the horizontal and vertical motion vectors of adjacent macroblocks is calculated, and if the difference is small, these two macroblocks are considered to represent the same object and are grouped together. The motion vector is encoded only once in the group. According to the method of the present invention, if it can be regarded as being equivalent to an adjacent block, it is assimilated, but it is possible not only to reduce error detection but also to reduce the code amount. The process of assimilating motion vectors is a means for the purpose, and the assimilated motion vectors are not encoded. Further, the method of the present invention can obtain the motion vector of the current block by tracing the direction of the block to be referred, and can reduce the code amount without increasing the calculation amount so much.

[0009]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the motion vector coding system of the present invention will be described in detail below with reference to FIGS. FIG. 1 shows an apparatus for realizing a motion vector coding method. This device is provided with a current frame memory 1, a macroblocking device 2, a motion vector generation device 3, a motion vector coding device 4, a search point output device 5, a previous frame memory 6, and a comparison macroblock generation device 7. There is.

The input digital image data is stored in the current frame memory 1. This current frame memory 1
The image data output from is supplied to the previous frame memory 6 and the macroblocking device 2. The macroblocking device 2 divides the supplied image data into a plurality of macroblocks. The comparison macroblock generation device 7 generates a comparison macroblock based on the search point input from the search point output device 5 and the previous frame image data output from the previous frame memory 6, and supplies it to the motion vector generation device 3. To do. In the motion vector generator 3,
A motion vector of the current macroblock is generated based on the macroblock data supplied from the macroblocking device 2 and the comparison macroblock supplied from the comparison macroblock generating device 7, and is output to the motion vector coding device 4. .

FIG. 2 shows an example of a group of macro blocks. When the motion vector obtained in the current macroblock can be regarded as equivalent to the motion vector in the adjacent macroblock, only the direction of the adjacent macroblock is coded and the motion vector is not coded. Macroblock comrades considered to be equivalent motion vectors
As shown by the diagonal lines in FIG. 2, the motion vectors are grouped and one motion vector is coded in the group. Each macroblock encodes only the direction of the macroblock to which it refers. For example, in order to encode three types of directions of upward, leftward and rightward, it is possible to perform encoding with only 2 bits.

FIG. 3 shows the order of comparison with adjacent macroblocks. As shown in this FIG.
The order of comparison with adjacent macroblocks is that the macroblock in the direction that started vertical scanning is first, the macroblock in the direction that started horizontal scanning is second, and that in the direction that ends horizontal scanning is The block is the third. Normally, scanning is performed from the upper left end of the frame to the right, so that the order of comparison is generally upper, left, and right as shown in FIG. The comparison ends when the motion vector of the current macroblock is considered equal to the motion vector of each comparison macroblock, and if the motion vector of the current macroblock is not equal in all three directions, the motion vector of the macroblock is encoded independently. .

Further, when referring to the adjacent block in the direction in which the scanning in the horizontal direction is finished, the motion vector of the outer adjacent block is determined and then referred to. That is, as shown in FIG. 4, for example, when determining the motion vector of the macroblock of mb1, mb1 should belong to the group g, but the motion vector of the macroblock mb2 to be referred to has not yet been determined. Therefore, first, the motion vector of mb2 is determined, and if the difference between the motion vectors of mb1 and mb2 can be considered to be equal, mb1 sets mb2 as the reference macroblock.

FIG. 5 shows an embodiment of grouping. When the surrounding macroblocks are grouped like mb3 in this figure and a motion vector independent of mb3 is determined, it is highly possible that mb3 belongs to the same group as the surroundings. In such a case, m
b3 is regarded as the same group as the surroundings and the same macroblock is referred to.

FIG. 6 shows a case of decoding a motion vector. When the motion vector is decoded on the decoding side, the macro blocks in the reference direction are sequentially traced, and when the macro block in which the motion vector exists is referred to, the motion vector of the block is set as the motion vector of the current macro block. For example, the mb of the group of FIG.
4 and mb5 follow macroblocks as shown by arrows, respectively, and finally, mb6 having independent motion vectors.
Is referred to as the motion vector of the current macroblock.

The method of this embodiment is based on ISO / IEC 13
It can also be used by incorporating it into the syntax of 818 (MPEG2). In the macro block layer syntax of MPEG2, some bits indicating the type of macro block are prepared. By adding the mode of the present embodiment to this, the system can be constructed without making a great change from the conventional system. For example, "macrob
"lock-type" is a variable-length code string that describes intra / inter, prediction direction, etc. However, if the mode of this embodiment is added to this, the code string has a long code length and the code amount cannot be reduced. "Frame-motion-type" or "field-m
“Otion-type” is a 2-bit code string that describes the motion prediction method. Although the three available modes are determined respectively, for example, the extended data for this method is read in the picture layer and “frame -motion-type "or" field-mot
If one of the "ion-type" modes is instructed to be replaced with this method, the same code structure as the conventional method can be adopted.

FIG. 8 shows an implementation example of MPEG2 syntax to which this method is applied. In this figure 8, referenc
e-direction is a flag when using this method, directio
The n-code is a code indicating the reference direction.

Further, FIG. 7 shows the flow of the operation for determining the motion vector of the macroblock by the method of this embodiment. First, the motion vector is determined (step 11), and the difference between the motion vector and the upward macroblock is evaluated (step 12). Then, it is determined whether the evaluation result is less than or equal to a predetermined threshold value (step 13). If it is less than or equal to a predetermined value (step 13; Y), the upward direction is determined as the reference direction (step 14), and the motion vector difference from the leftward macroblock is evaluated (step 15). If the evaluation value is less than or equal to the threshold value (step 16; Y), the reference direction of the macro block in the left direction is determined to be the right direction (step 17), and if it is not less than or equal to the threshold value (step 16; N),
The macroblock to the left has an independent motion vector (step 18).

On the other hand, in step 13, if the evaluation value is not less than the threshold value (N), the difference between the motion vector and the macro block in the left direction is evaluated (step 19), and whether the evaluation value is less than the threshold value again. Is determined (step 20). Here, when it is less than or equal to the threshold value (step 20; Y), the reference direction of the macro block in the left direction is determined as the right direction (step 21), and the left direction is determined as the reference direction (step 2).
2). On the other hand, if it is not equal to or less than the threshold value in step 20 (N), the code is determined by the next macroblock (step 23).

As described above, according to the motion vector coding / decoding method of this embodiment, the code amount of a moving image can be reduced. In addition, compatibility can be maintained even when incorporated in the conventional MPEG system syntax. When this method is used, it is expected that the decoding side will take more time than before in decoding the motion vector.
However, when comparing the transmission speed of the current general wire / wireless transmission path and the processing speed of the arithmetic unit of the decoder, the processing speed of the decoder is generally much higher, and the decoder is somewhat burdened. It is more meaningful for real-time transmission to lighten the burden on the transmission path even if the load is forced.

[0021]

According to the present invention, it is possible to reduce the information amount of moving image data. In addition, compatibility can be maintained even when incorporated in the conventional MPEG syntax. Further, it is possible to reduce the code amount while suppressing the increase in the calculation amount.

[Brief description of drawings]

FIG. 1 is a block configuration diagram of an apparatus that implements a motion vector encoding method according to an embodiment of the present invention.

FIG. 2 is an explanatory diagram showing an example of a group of macro blocks.

FIG. 3 is an explanatory diagram of an order of comparison with adjacent macroblocks.

FIG. 4 is an explanatory diagram for referring to the adjacent block in a direction in which horizontal scanning ends.

FIG. 5 is an explanatory diagram regarding grouping.

FIG. 6 is an explanatory diagram for decoding a motion vector.

FIG. 7 is a flowchart showing an operation of determining a motion vector of a macro block.

FIG. 8 is an explanatory diagram showing an implementation example of MPEG2 syntax to which the present system is applied.

[Explanation of symbols]

 1 Current Frame Memory 2 Macroblocking Device 3 Motion Vector Generating Device 4 Motion Vector Encoding Device 5 Search Point Output Device 6 Previous Frame Memory 7 Comparison Macroblock Generating Device

Claims (4)

[Claims] 1. An input digital video signal is divided into a plurality of blocks, the blocks are sequentially scanned, each block is subjected to block matching with a previous frame to detect a motion vector, and the obtained motion vector is calculated. In the method of transmitting together with the pixel information of an image, the comparison judgment means for comparing the motion vector obtained in the current block with the motion vector in the adjacent block and judging whether both the motion vectors can be regarded as equivalent, and this comparison A motion vector coding system, comprising: a coding unit that, when judged to be equivalent by the judging unit, does not code the motion vector but codes the direction of the adjacent block. 2. When comparing with an adjacent block, the comparison / determination means sets a block in a direction in which vertical scanning is started as a first block, a block in a direction in which horizontal scanning is started as a second block, and a horizontal direction. 2. The block in the direction in which the scanning of 3 is finished is compared as the third block, and the comparison is performed.
The described motion vector coding method. 3. The comparison / determination means, in the case of performing comparison / determination with an adjacent block in a direction in which horizontal scanning is completed, makes a comparison / determination after determining a motion vector of the adjacent block. The motion vector coding method according to claim 2. 4. A receiving means for receiving a coded motion vector and the direction of an adjacent block, and a block in the reference direction are sequentially traced from the block of the direction received by this receiving means, and a motion vector exists. A motion vector decoding system, comprising: a decoding unit that, when a block is referred to, decodes the motion vector of the block as a motion vector of the current block.