JP2537244B2 - Motion compensation motion vector coding method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Outline] A motion vector coding method that incorporates motion compensation in a motion picture screen, and a predetermined number of input screen blocks for the purpose of reducing the amount of information transmitted by the motion compensation motion vector. A codebook that stores a plurality of prediction error patterns composed of prediction error motion vectors, and a prediction error vector pattern that generates a prediction motion vector pattern corresponding to the input screen block. A predictor which is stored as an index and the previous screen stored by inputting the predicted motion vector pattern and stored in the frame memory are motion-compensated and compared with the input screen, and the predicted motion vector pattern with the smallest error is selected and the predicted motion vector pattern is selected. Matching that sends the index of the prediction error pattern corresponding to the vector pattern to the receiving side via the entropy coder A predicting motion vector pattern from the previous predictive motion vector pattern and a prediction error pattern obtained by decoding the index output from the matching determining circuit with a decoder. To generate.

[Industrial applications]

The present invention relates to an image coding system, and more particularly to a motion vector coding system that incorporates motion compensation in a moving image screen.

In recent image transmission such as video conferencing, various proposals have been made from the viewpoint of realizing a low bit rate by compressing the transmission amount as much as possible in order to effectively use the transmission line and reduce the terminal cost. However, the so-called block matching type motion compensation method, which detects the motion of the transmission screen and obtains the difference signal by compensating the prediction screen by the amount of the motion and further quantizing and transmitting the difference signal, is in the limelight. Is coming.

[Conventional technology]

FIG. 11 shows a well-known coding system with motion compensation, which was announced at the Institute of Electronics and Communication Engineers on April 5, 1984, for example.

In the figure, 11 is a quantizer, 12 is a frame memory, 13 is a variable delay device, and 14 is a motion detector. In this method, as shown in FIG. 12, input screen block x (I, J) Absolute value | x (I, J)-(y (i, j) of the difference in pixel units between the same position on the reference screen stored in the frame memory 12 and the surrounding block y (i, j). ) | Is taken by the motion detector 14 and accumulated for one block, and the accumulated value Block is predicted to be the closest to the screen block before moving, that is, the predicted block is delayed by the amount moved by the variable delay unit 13 and the difference with the next input screen block is obtained. The quantizer 11 quantizes and sends out to the transmission path. At the same time, the prediction screen is stored in the frame memory 12 (via an inverse quantizer (not shown)) and is used to determine the next prediction block. Then, the detected amount of motion of the block (hereinafter, referred to as a motion vector value) is separated from (or together with) the coding circuit of the quantized output of the quantizer 11 by the entropy coder (coding circuit) shown in FIG. ) Send to receiver via 10.

The receiving side reproduces the original screen from the received signal according to the motion detection information.

[Problems to be solved by the invention]

In the above conventional motion compensation motion vector entropy coder as shown in FIG. 13, a set of a motion vector in the X-axis (vertical axis) direction and a motion vector in the Y-axis (horizontal axis) direction (V _(X , V _Y ) is appropriately variable-length coded and transmitted individually to the receiving side, so although individual motion vectors can be transmitted accurately, the smaller the size of the block when performing motion compensation becomes The problem is that the amount of information increases due to motion compensation.

Therefore, it is an object of the present invention to reduce the amount of information transmitted by a motion compensation vector.

[Means for solving problems]

FIG. 1 conceptually shows a motion compensation motion vector encoding system according to the present invention for achieving the above object. 1 is composed of prediction error motion vectors corresponding to a predetermined number of input screen blocks. A codebook that stores a plurality of prediction error patterns, 2 is a predictor that generates a prediction motion vector pattern corresponding to the input screen block, and stores a prediction error pattern that corresponds to the prediction motion vector pattern as an index, 3 Inputs the predicted motion vector pattern, compares the previous screen stored in the frame memory 12 with the input screen as motion compensation, selects the predicted motion vector pattern with the smallest error, and selects the prediction error corresponding to the predicted motion vector pattern. Entropy coder for pattern index 10
The predictor 2 is a matching determination circuit that is sent to the receiving side via the predictor 2 and the prediction error obtained by decoding the previous prediction motion vector pattern and the index output from the matching determination circuit 3 by the decoder 4. The predicted motion vector pattern is generated from the pattern.

[Action]

In the motion compensation motion vector encoding method of the present invention shown in FIG. 1, a plurality of prediction error patterns (prediction error motion vectors) composed of prediction error motion vectors corresponding to a predetermined number of input screen blocks are included in the codebook 1. Pattern) is stored. The matching determination circuit 3 inputs every predetermined number of input screen blocks, and also inputs the predicted motion vector pattern output from the predictor 2 corresponding to these input screen blocks. This predicted motion vector pattern is
It is obtained from the prediction error pattern obtained by decoding the index output from the matching determination circuit 3 by the decoder 4 and the previous prediction motion vector pattern stored in the predictor 2.

The matching determination circuit 3 motion-compensates the previous screen block from the frame memory 12 according to the predictive motion vector pattern thus obtained, and compares the input screen block with the previous screen block. Then, as a result of the comparison, the predicted motion vector pattern having the smallest error, that is, the distortion is selected, and in order to obtain the prediction error pattern corresponding to the predicted motion vector pattern, the selected prediction motion vector pattern is sent to the predictor 2. The index of the corresponding prediction error pattern in the codebook 1 stored in the predictor 2 is read and the index is used as an entropy coder.
Send to 10. The entropy coder 10 encodes the index into a variable length code, sends it to the transmission path, and sends it to the receiving side.

In this way, the amount of information is compressed and transmitted in the form of the index of the motion vector pattern in which a predetermined number of screen blocks are collectively matched, without individually compensating the motion of the input screen blocks.

〔Example〕

Embodiments of the motion compensation motion vector encoding method according to the present invention will be described below.

FIG. 2 shows an embodiment in which the motion compensation motion vector coding system of the present invention shown in FIG. 1 is incorporated into a motion compensation coding system. In this embodiment, the motion detector is Instead, the matching determination circuit 3 performs the function instead.

That is, the matching determination circuit 3 obtains the prediction error (motion vector) pattern stored in the codebook 1 by the decoder 4 using the transmission index, and the predictor 2 calculates the prediction motion vector pattern (this is the previous prediction motion vector pattern). And a prediction error pattern are added) to obtain a previous screen (prediction screen) stored in the frame memory 12
Motion compensation.

An example of this motion compensation is shown in FIG. 3, and in this embodiment, the previous screen of the screen block corresponding to the number (4) of motion vectors of the motion vector pattern is motion compensated and the input screen is Input the same number of blocks and compare both.

By this comparison, the prediction motion vector pattern having the smallest error is selected, and the matching determination circuit 3 inquires of the predictor 2 which prediction error pattern of the codebook 1 this prediction motion vector pattern corresponds to. Predictor 2
Stores the index of the prediction error pattern used when obtaining the prediction motion vector pattern, and sends the index of the prediction error pattern to the matching determination circuit 3 by the above inquiry. Matching determination circuit 3
Transmits the index to the receiving side via the entropy coder 10.

The selected predicted motion vector pattern is also given to the variable delay unit 13. As a result, the variable delay unit 13 similarly shifts the previous screen block in the frame memory 12 to generate a predicted screen.

Since various examples can be considered for the motion vector pattern stored in the codebook 1, those examples will be described below.

FIG. 4 shows a prediction error motion vector pattern stored in the codebook 1 in a pair in the X-axis direction and the Y-axis direction. The prediction error motion vector pattern is formed by n prediction error motion vector patterns. There are m codes (this corresponds to the number of indexes). In the example of FIG. 3, n = 4, and motion compensation is individually performed on four screen blocks.

An embodiment of the judgment operation in the matching judgment circuit 3 based on this prediction error motion vector pattern will be described with reference to the flowchart shown in FIG.

First, each predictive motion vector code of the predictive motion vector pattern generated by the predictor 2 and input to the matching determination circuit 3 is C ′ (1,1) when n = 3 and m = 2 in FIG. ) = C (1,1) + V _x C '(2,1) = C (2,1) + C' (1,1) C '(3,1) = C (3,1) + C' (2, 1) C '(1,2) = C (1,2) + V _y C' (2,2) = C (2,2) + C '(1,2) C' (3,2) = C (3 , 2) + C '(2,2). However, (V _x , V _y ) was stored in the predictor 2 1
This is the previous predicted motion vector value and is C (1,1) to C (3,1)
Is the X-axis direction value of the prediction error motion vector value of the prediction error motion vector pattern, and C (1,2) to C (3,2) is the Y axis direction value of the prediction error motion vector value of the prediction error motion vector pattern. , Code C '(1,1) with dash' added to them ~
C '(3,2) indicates the predicted motion vector of the predicted motion vector pattern in the X-axis direction and the Y-axis direction, respectively. Predictor 2
In, the prediction error pattern is stored in the form of the corresponding index.

As shown in FIG. 3, the input screen is divided into n-sized blocks for motion compensation.
Let F ₁ (f _x1 , f _y1 ), F ₂ (f _x2 , f _y2 ), ..., F _n (f _xn , f _yn ). Further, the block previous screen from the frame memory 12 is P (x, y), and the prediction motion vector of the prediction motion vector pattern is a block as shown in FIG.
(N, m), variables for determination are R, IDX (index), D ₁ , D ₂ , ... D _n , D _s (errors for f ₁ , f ₂ , ..., f _n , respectively, D _s is (Total error) (step S in FIG. 7)
1).

As an initial value, R = ∞ (however, ∞ means the maximum value that the cumulative value D _s can take), J (count) = 0, D = 0, I
= 0 (step S2). Then, in step S3, J
Is incremented by "1".

In step S4, I is incremented by "1", and in step S5, the previous screen P (x, y), which is blocked in the same manner as the input screen, is predicted by the motion vector code C '(I, J).
Motion compensation is performed by (I = 1 to n, J = 1 to m) to obtain a block screen P ′ (x, y) (at step S5).

Then, the absolute value error (represented by ABS) D _I between the motion-compensated block screen P ′ (x, y) and the input block screen F _I (f _xI , f _yI ) is obtained (at step S6), and this is obtained. The value obtained by adding D _s is set as a new D _s (at step S7), and steps S4 to S8 are repeated until I reaches n. As a result, as shown in FIG. 4, 1 to 1 of the motion vector pattern of index "1"
The error between each predicted motion vector value of n and the X-axis direction vector (or Y-axis direction vector) is accumulated.

When I reaches n and index "1" ends, R
And D _s are compared (the same step S9), and when D _s has not reached R, R = D _s and the current cumulative value D _{s is set} to the previous cumulative value.
D _s and IDX = J (step S10), and steps S3 to S11 are repeated until J = m. This gives m
The error is accumulated for each of the predictive motion vector patterns, and steps S9 and S10 reduce the D
_{Since the} IDX that gives _s is obtained, the predicted motion vector pattern that gives the smallest error (distortion) is finally selected (step S12).

The predictive motion vector pattern selected in this way is
In the future, it is necessary to find out which one of the prediction error patterns in Codebook 1 corresponds.

Therefore, the predictor 2 inputs the prediction error motion vector pattern (see FIG. 4) to be added to the previous prediction motion vector pattern in order to generate the prediction motion vector pattern from the codebook 1 corresponding to the input screen block. Then, the index corresponding to each prediction error pattern is stored.

Therefore, the matching determination circuit 3 queries the predictor 2 for the index, reads the prediction error pattern and the index (at step S13), and sends the index (at step S14) to the entropy coder 10. It will be.

In step S6 for _{obtaining the} error, the square error {P '(x, y) -F _I (f _XI , f _yI )} ² or the average value may be used instead of the absolute value.

FIG. 6 shows another embodiment of the codebook 1. For the X-axis direction, there are m _x (1 to m) prediction error motion vector patterns formed by collecting n _x prediction error motion vector values. _{x is used} as an index of each prediction error motion vector pattern), and the prediction error motion vector pattern in which _{y y} prediction error motion vector values are collected is m
_y (each of 1 to m _y serves as an index of the prediction error motion vector pattern) are prepared. These predictions erroneous differential vector pattern stored in advance obtained in advance but, n _x = n _y, even _{m x = m y, n x} ≠ n y, m x ≠
It may be m _y . However, in the case of n _x = n _y , for X-axis direction and Y
It is not necessary to separate codebook 2 for the axial direction,
In that case, the pattern configuration is the same as that shown in FIG.

An embodiment (n _x = n _y ) of the operation of the matching determination circuit 2 according to the configuration example of such a codebook is shown in the flowchart of FIG. For axial direction C _x (n _x , m
_x ) and C _y (n _y , m _y ) are used for the Y-axis direction, except that C _y (n _y , m _y ) is used. However, variables K and J are used to search the numbers m _x and m _y of the prediction error motion vector patterns.

Although n _x = n _y is set in the embodiment of FIG. 8, if n _x ≠ n _y , the number n of blocks subject to motion compensation is
n _x, when divisible by greater than n _y and n _x n _y, for example by using a codebook concatenating code as shown in FIG. 9, X-axis direction for each n _x pieces, Y-axis direction is n Motion compensation can be performed for each block using the same prediction error motion vector for every _y .

Furthermore, since the motion vector values of adjacent blocks are relatively similar to each other, the distortion for the code in the codebook is not calculated as above, but the code is searched in a tree structure as shown in FIG. You may.

On the other hand, on the receiving side, the reverse operation is performed to reproduce the motion vector.

That is, the prediction error motion vector pattern (this is the same codebook as the transmitter is prepared on the receiver side) is extracted by the transmitted index, and the number of dimensions (prediction error motion vector pattern) of this prediction error motion vector pattern is extracted. It is sufficient to perform motion compensation using the same prediction error motion vector pattern for the reproduction blocks corresponding to the number n) of prediction error motion vectors of the vector pattern.

Therefore, only the index of the matched motion vector pattern needs to be sent from the transmitting side to the receiving side.

〔The invention's effect〕

As described above, according to the motion compensation vector encoding method of the present invention, a plurality of prediction error motion vector patterns are prepared corresponding to the screen blocks, and the prediction error motion vector patterns are converted into the prediction motion vector patterns. Since it is configured to compare the converted and motion-compensated screen with the input screen and select the prediction error motion vector pattern with the minimum distortion and transmit the index, the predetermined number of input screen blocks is 1 This can be represented by one prediction error motion vector pattern, and although some distortion occurs, the amount of information is small and extremely efficient coded transmission can be realized.

[Brief description of drawings]

FIG. 1 is a block diagram showing the principle of a motion compensation motion vector coding system according to the present invention, and FIG. 2 is an embodiment of a motion compensation coding system incorporating the motion compensation motion vector coding system according to the present invention. FIG. 3 is a diagram for explaining block motion compensation, FIG. 4 is a diagram showing an embodiment of a codebook used in the present invention, and FIG. 5 is an embodiment of the operation of a matching determination circuit. 6 and 7 are diagrams showing an embodiment of a code book used in the present invention, and FIG. 8 is a flowchart for explaining another embodiment of the operation of the matching judgment circuit. Fig. 9 is a diagram showing the concatenation of the codes of the codebook used when the number of prediction error motion vectors in the X-axis direction and the Y-axis direction are different. Fig. 10 is a diagram showing the tree-structure codebook. Figure is known FIG. 12 is a block diagram showing an example of the motion compensation coding method adopted, FIG. 12 is a diagram for explaining the principle of motion compensation, and FIG. 13 is a diagram showing a conventional entropy coder for coding a motion vector. In FIG. 1, 1 ... Codebook, 2 ... Predictor, 3 ... Matching decision circuit, 4 ... Decoder, 10 ... Entropy coder 12 ... Frame memory. In the drawings, the same reference numerals indicate the same or corresponding parts.

Claims (4)

(57) [Claims] 1. A codebook (1) storing a plurality of prediction error patterns composed of prediction error motion vectors corresponding to a predetermined number of input screen blocks, and a prediction motion vector pattern corresponding to the input screen blocks. A predictor (2) that stores a prediction error pattern generated and corresponding to the predicted motion vector pattern as an index, and a predictor (2) that inputs the predicted motion vector pattern
2) The previous screen described in 2) is motion-compensated and compared with the input screen, the prediction motion vector pattern with the smallest error is selected, and the index of the prediction error pattern corresponding to the prediction motion vector pattern is set to the entropy coder (10). A matching determination circuit (3) to be transmitted to the receiving side via the predictor (2), and the predictor (2) decodes the previous predictive motion vector pattern and the index output from the matching determination circuit (3) into a decoder. A motion compensation motion vector coding method characterized in that the prediction motion vector pattern is generated from the prediction error pattern obtained by decoding in (4). 2. The codebook (1) has a Y-axis direction and a Y-axis direction.
The motion-compensated motion vector coding method according to claim 1, wherein the prediction error pattern is stored separately in the axial direction. 3. The motion compensation motion vector coding method according to claim 2, wherein the number of each value of the prediction error patterns divided in the X-axis direction and the Y-axis direction is different. 4. The code book (1) has a Y-axis direction and a Y-axis direction.
The motion compensation motion vector encoding method according to claim 2, wherein the prediction error patterns divided in the axial direction are stored in pairs.