JPH0918882A - Motion vector search method and apparatus thereof - Google Patents

Abstract

(57) [Abstract] [Purpose] An object of the present invention is to arbitrarily reduce a calculation amount for motion vector search in a motion vector search method of a motion compensation interframe coding method. An input frame memory 1a in which a video input signal is stored, a reference frame memory 1b in which a previous frame video signal is stored, and a sub-sampling unit 1c are provided, and a sub-sampling pattern of the sub-sampling unit 1c is controlled by a sub-sampling control unit 1g. By controlling, the current block data and the reference block data are fetched and sub-sampled, and the two types of sub-sampling data are evaluated and calculated by the evaluation function calculation unit 1d. The threshold value comparing unit 1e compares the evaluation function value during the calculation to determine whether the preceding evaluation function calculation is significant. When it is determined to be significant, the evaluation function calculation is continued, and it is determined to be insignificant. At this time, the evaluation function calculation is aborted, and the evaluation function value being calculated is output to the motion vector determination unit 1f.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a motion vector search method and device in a motion compensation interframe coding system in the field of moving image signal processing.

[0002]

2. Description of the Related Art As one of the methods for compressing a digital moving image signal, there is an interframe coding method which takes a difference between adjacent frames. In this method, the compression rate is high when the movement between frames is small, but when the movement is large, the difference becomes large and the compression rate decreases.

The motion-compensated inter-frame coding system was conceived there. This is a method of detecting a motion vector, moving the previous frame by the motion vector, and then searching for a difference from the current frame. As a motion vector search method in the motion-compensated interframe coding method, there are generally the following conventional examples.

FIG. 4 shows the structure of a conventional motion vector search method and apparatus. In FIG. 4, reference numeral 4a is an input frame memory for accumulating a video input signal, and sends the data of the current block to the sub-sampling unit 4c.
Reference numeral 4b is a reference frame memory for accumulating the previous frame signal, and sends the data of the reference block to the sub-sampling unit 4c. Reference numeral 4d is an evaluation function calculation unit that calculates an evaluation function, and sends the calculation result to the motion vector determination unit 4f. The motion vector determination unit 4f is the sub-sampling unit 4
It controls which reference block is selected for c and outputs a motion vector.

The sub-sampling unit 4c, the evaluation relation calculating unit 4d and the motion vector judging unit 4f constitute a motion compensating unit.

Next, the operation of the above conventional example will be described.
In FIG. 4, the video input signal is the input frame memory 4a.
Sent to The reference frame memory 4b stores the previous frame video signal. The sub-sampling unit 4c sub-samples the current block data from 4a and the reference block data from the reference frame memory 4b with a fixed sampling pattern. The two types of sub-sampling data are sent to the evaluation function calculation unit 4d, and the evaluation function is calculated there. A sum of absolute values of difference signals or a sum of squared values of difference signals is often used as the evaluation function. The result of this evaluation function value is sent to the motion vector determination unit 4f. The motion vector determination unit 4f checks whether or not all the search range has been searched, and when the search is not completed yet, the sub-sampling unit 4c is controlled to change the reference block data and continue the search, When the search is completed, the motion vector is determined and output.

As described above, in the above-described conventional motion vector search method, the sub-sampling unit sub-samples the current block data and the reference block data with a predetermined sampling pattern, and the evaluation function calculation unit calculates the evaluation function. The motion vector can be searched.

[0008]

However, in the above-described conventional motion vector search method, since the sampling pattern is fixed, the sub-sampling is performed in order to further reduce the calculation amount of the evaluation function and conversely increase the accuracy of the motion vector. There was a problem that could not be dense. Furthermore, since the evaluation function calculation is performed on all sampling data, even if it is known that the evaluation function value has already become a large value during calculation and continuation of the evaluation function calculation is unnecessary for motion vector search, the calculation is performed. Since it is executed, there is a problem that the amount of calculation is large.

Further, in the above-described conventional motion vector search method, since the encoder including the motion compensation unit is assumed to be a system independent of the decoder, the image coding apparatus and the image decoding apparatus have a certain processing capability. Moreover, when it is attempted to realize it on a single arithmetic unit (for example, a CPU), the processing amount of other image encoding units and decoding units is limited due to the large amount of processing amount of motion vector search, resulting in There was a problem that the image quality deteriorates.

The present invention solves such a conventional problem, and an object thereof is to provide an excellent motion vector search method capable of arbitrarily reducing the amount of calculation for motion vector search. Is.

[0011]

In order to achieve the above object, the present invention provides a motion vector detection method of a motion compensation interframe coding method, which is a subsampling unit for subsampling the current block data and reference block data. Is provided with a sub-sampling pattern control unit for controlling the sampling pattern, and pixel data of an evaluation function used at the time of motion vector search is variably sub-sampled.

Further, according to the present invention, there is provided a threshold value comparing section for the evaluation function value being used for calculating the evaluation function used at the time of motion vector search, and the evaluation function operation being calculated by the threshold value comparing section is preceded. It is characterized in that it is judged whether or not doing is significant, and when it is judged not significant, the calculation of the evaluation function is terminated.

Further, the present invention has a receiving buffer for storing the coded video signal of the decoding unit, and the sub-sampling unit according to the residual amount of the coded video signal of the receiving buffer processed by the decoding unit. The sub-sampling pattern of is controlled.

[0014]

Therefore, according to the present invention, the data used for the evaluation function can be variably sub-sampled by the sub-sampling control unit, and the amount of calculation for motion vector search can be arbitrarily reduced.

Further, according to the present invention, by applying the threshold value comparing unit to the evaluation function value during calculation, it is possible to reduce the insignificant evaluation function operations and reduce the operation amount for the motion vector search. You can

Further, according to the present invention, the encoding / decoding device by the single arithmetic unit (CPU) is realized by controlling the sampling pattern according to the size of the residual amount in the reception buffer for decoding processing.

[0017]

1 is a block diagram showing a first embodiment of a motion vector search method according to the present invention.

In FIG. 1, 1a is an input frame memory for storing a video input signal, and 1b is a reference frame memory for storing a previous frame signal. Reference numeral 1c denotes a sub-sampling unit that takes in data used for the evaluation function from the input frame memory 1a and the reference frame memory 1b, and the sub-sampling control unit 1g controls the sampling pattern of the sub-sampling unit 1c. 1d is an evaluation function calculation unit that calculates an evaluation function based on the subsampled data, and the evaluation function value being calculated is the threshold comparison unit 1e.
And the result of comparison of the evaluation function is sent to the motion vector determination unit 1f. The motion vector determination unit 1f determines whether all search ranges have been searched, determines a motion vector, and outputs the motion vector.

The sub-sampling unit 1c, the evaluation function calculating unit 1d, the threshold comparing unit 1e and the motion vector determining unit 1f constitute a motion compensating unit.

Next, the operation of the first embodiment constructed as described above will be explained. In this embodiment, the video input signal is stored in the input frame memory 1a. The reference frame memory 1b stores the data of the previous frame. This previous frame data is generally generated by decoding by the local decoding unit which is a part of the encoder. The sub-sampling unit 1c sub-samples the pixel data used for the evaluation function, and its sampling pattern is determined by the sub-sampling control unit 1g.

FIGS. 2A to 2C show an example of the sampling pattern determined by the sub-sampling control unit 1g in the first embodiment. In a block of 8 × 8 pixels, a circle is marked. Sampled pixels.

The sub-sampling section 1c sub-samples the pixel data of the current block from the input frame memory 1a with the determined sampling pattern. further,
The sub-sampling unit 1c determines which reference block to compare with the current block based on the block at the same position in the previous frame, and sub-samples the pixel data of the reference block from the reference frame memory 1b with the determined sampling pattern. Both data obtained by the sub-sampling unit 1c are passed to the evaluation function calculation unit 1d to calculate the evaluation function value, and the evaluation function value being calculated is set to the threshold comparison unit 1
e. In the threshold value comparison unit 1e, the evaluation function being calculated is compared with the threshold value, and if the calculation value exceeds the threshold value, the calculation of the evaluation function is terminated, and the evaluation function value being calculated is determined as the motion vector determination unit 1f. Output to. When the evaluation function value being calculated does not exceed the threshold value, the evaluation function calculation unit 1d is controlled to continue the evaluation function calculation.

The evaluation function value thus obtained is sent to the motion vector determination section 1f. The motion vector determination unit 1f determines whether or not all search ranges have been searched. When all the search ranges have not been searched yet, the sub-sampling unit 1c is controlled to change the reference block data and continue the search. When the search for all the search ranges is completed, the minimum evaluation function value is set. To detect. The difference between the position of the reference block and the position of the current block when this minimum evaluation function value is given is obtained as a motion vector, and the obtained motion vector is output.

As described above, according to the first embodiment,
By providing the sub-sampling control unit, the sub-sampling unit can variably sub-sample pixel data used for the evaluation function, and has an advantage that the amount of calculation for motion vector search can be arbitrarily reduced.

Further, according to the first embodiment, by providing the threshold value comparing section for the evaluation function value being calculated, the evaluation function calculating section can abort the calculation of the insignificant evaluation function on the way. This has the advantage that the amount of calculation for motion vector search can be reduced.

FIG. 3 is a block diagram showing a second embodiment of the motion vector search method according to the present invention.

In FIG. 3, reference numeral 20 is an encoding / decoding device composed of a single arithmetic unit, 2a is an input frame memory for storing a video input signal, and 2b is a reference frame memory for storing a previous frame signal. 2c is the input frame memory 2a and the reference frame memory 2b for using the data used for the evaluation function
The sub-sampling control unit 2g controls the sampling pattern of the sub-sampling unit 2c.

Reference numeral 2d is an evaluation function calculation unit for calculating an evaluation function based on the sub-sampled data, the evaluation function value being calculated is sent to the threshold value comparison unit 2e, and the comparison result of the evaluation function is It is sent to the motion vector determination unit 2f. Reference numeral 2f is a motion vector determination unit that determines whether or not all search ranges have been searched and determines a motion vector, and outputs the motion vector to another encoding unit 2h. Other coding unit 2h
Is for coding other than motion compensation, and outputs a coded video output signal. Reference numeral 2i is a local decoding unit that generates data to be stored in the reference frame memory 2b, and 2j is a reception buffer, which stores an encoded video input signal. 2k
Is a decoding unit that outputs a video output signal.

The sub-sampling unit 2c, the evaluation function calculating unit 2d, the threshold comparing unit 2e and the motion vector judging unit 2f constitute a motion compensating unit.

Next, the operation of the second embodiment constructed as described above will be explained. In this embodiment, the video input signal is stored in the input frame memory 2a, and the reference frame memory 2b stores the data of the previous frame.
The previous frame data is generated by decoding the coded video output signal coded by the other coder unit 2h by the local decoding unit 2i.

On the other hand, the coded video input signal is stored in the reception buffer 2j, and the reception buffer residual amount is sent to the sub-sampling control unit 2g. Sub-sampling control unit 2
g controls the sampling pattern for the sub-sampling unit 2c based on this information. Further, the encoded video input signal is decoded by the decoding unit 2k through the reception buffer 2j and output as a video output signal.

The sub-sampling unit 2c sub-samples the pixel data used for the evaluation function from the input frame memory 2a and the reference frame memory 2b. The sampling pattern is determined by the control of the sub-sampling control unit 2g.

An example of the sampling pattern determined by the sub-sampling control unit 2g in the second embodiment is shown in FIG. 2, in which a pixel with a circle is sampled in a block of 8 × 8 pixels. Further, in the second embodiment, the residual amount of the receiving buffer is used as a judgment material for the sampling pattern control.

Specifically, the sub-sampling unit 2c is controlled to increase the sampling data if the remaining amount of the receiving buffer is large and decrease the sampling data if the remaining amount of the receiving buffer is small. Then, based on the sampling pattern, the pixel data of the current block is sub-sampled from the input frame memory 2a. Further, the sub-sampling unit 2c determines which reference block to compare with the current block based on the block at the same position in the previous frame, and sub-pixels the reference block pixel data from the reference frame memory 2b with the determined sampling pattern. To sample. Sub-sampling unit 2
Both data obtained in c are passed to the evaluation function calculation unit 2d to calculate the evaluation function value, and the evaluation function value being calculated is output to the threshold value comparison unit 2e. In the threshold comparison unit 2e,
The evaluation function value being calculated is compared with the threshold value, and if the value exceeds the threshold value, the calculation of the evaluation function is aborted, regardless of whether the calculation is in progress.
The evaluation function value being calculated is output to the motion vector determination unit 2f. If the evaluation function value being calculated does not exceed the threshold value, the evaluation function calculation unit 2d is controlled to continue the evaluation function calculation.

The evaluation function value thus obtained is sent to the motion vector determination section 2f. In 2f, it is determined whether or not all the search ranges have been searched. If all the search ranges have not been searched yet, the sub-sampling unit 2c
The reference block data is changed to control to continue the search, and when the search in all search ranges is completed, the minimum evaluation function value is detected. The difference between the position of the reference block and the position of the current block when the minimum evaluation function value is given is obtained as a motion vector, and the obtained motion vector is output to another encoding unit 2h.

As described above, according to the second embodiment, by providing the sub-sampling control section controlled by the reception buffer, the sub-sampling section can variably sub-sample pixel data used for the evaluation function. The advantage is that the amount of calculation for motion vector search can be arbitrarily reduced.

Further, according to the second embodiment, by providing the threshold value comparing section for the evaluation function value being calculated,
When the precedence of the evaluation function calculation is not significant, the calculation of the threshold evaluation function can be stopped midway, and the amount of calculation for motion vector search can be reduced.

Further, according to the second embodiment, the sub-sampling control unit controlled according to the remaining amount of the reception buffer is provided on the encoding / decoding device composed of the single arithmetic unit (CPU). The sub-sampling unit can variably sub-sample the pixel data used for the evaluation function, and the load of the encoding processing unit is variably changed according to the size of the processing of the decoding processing unit, so that a stable single arithmetic unit can be used. This has the effect of realizing an encoding / decoding device.

[0039]

As is apparent from the above-described embodiments of the present invention, by providing the sub-sampling control unit in the motion vector search method, the data used by the sub-sampling unit for the evaluation function can be variably sub-sampled. The advantage is that the calculation amount for motion vector search can be arbitrarily reduced.

Further, according to the present invention, in the motion vector search method, by providing the threshold value comparison unit for the evaluation function value being calculated, the evaluation function calculation unit aborts the insignificant evaluation function calculation even during calculation. Therefore, there is an effect that the calculation amount for motion vector search can be reduced.

Further, according to the present invention, in the motion vector search method, by providing the sub-sampling control section controlled by the reception buffer, the pixel data used by the sub-sampling section for the evaluation function is used as the processing amount of the decoding section. Since variable sub-sampling is performed accordingly, the load of the encoding processing unit is variably changed depending on the size of the processing of the decoding processing unit, so that the encoding / decoding device can be realized on a stable single arithmetic unit. Have.

[Brief description of the drawings]

FIG. 1 is a schematic block diagram showing a configuration of a motion vector search method and apparatus according to a first embodiment of the present invention.

FIG. 2 is a diagram showing an example of a sampling pattern in the embodiment of the present invention.

FIG. 3 is a schematic block diagram showing a configuration of a motion vector search method and apparatus according to a second embodiment of the present invention.

FIG. 4 is a schematic block diagram showing a configuration of a conventional motion vector search method and apparatus.

[Explanation of symbols]

 1a, 2a, 4a Input frame memory 1b, 2b, 4b Reference frame memory 1c, 2c, 4c Sub-sampling unit 1d, 2d, 4d Evaluation function calculation unit 1e, 2e Threshold value comparison unit 1f, 2f, 4f Motion vector determination unit 1g, 2g Sub-sampling control unit 2h Other encoder unit 2i Local decoding unit 2j Receive buffer 2k Decoder unit 20 Encoding / decoding device on single arithmetic unit

Claims (6)

[Claims] 1. A motion vector detection method using a motion-compensated inter-frame coding method, comprising: a sub-sampling pattern control unit that controls a sampling pattern of a sub-sampling unit that sub-samples current block data and reference block data. A motion vector search method characterized in that pixel data of an evaluation function used during vector search is variably subsampled. 2. It is significant that a threshold value comparison unit is provided for an evaluation function value being calculated for use in the motion vector search, and the evaluation function calculation being calculated is preceded by this threshold value comparison unit. The motion vector search method according to claim 1, wherein it is judged whether or not it is judged, and when it is judged not significant, the calculation of the evaluation function is terminated. 3. A sub-sampling pattern of the sub-sampling unit, the sub-sampling pattern having a receiving buffer for storing the encoded video signal of the decoding unit, the sub-sampling unit according to the residual amount of the encoded video signal of the receiving buffer processed by the decoding unit. 3. The motion vector search method according to claim 1, further comprising: 4. A motion vector detection method using a motion-compensated inter-frame coding method, comprising: a sub-sampling pattern control unit that controls a sampling pattern of a sub-sampling unit that sub-samples current block data and reference block data. A motion vector search device characterized in that pixel data of an evaluation function used in vector search is variably subsampled. 5. It is significant that a threshold value comparison unit is provided for an evaluation function value being calculated for use in the motion vector search, and the evaluation function operation being calculated is preceded by this threshold value comparison unit. 5. The motion vector search device according to claim 4, wherein it is determined whether or not it is determined, and when it is determined that it is not significant, the calculation of the evaluation function is terminated. 6. A sub-sampling pattern of the sub-sampling unit, the sub-sampling pattern having a receiving buffer for storing the encoded video signal of the decoding unit, and the sub-sampling unit according to the residual amount of the encoded video signal of the receiving buffer processed by the decoding unit 6. The motion vector search device according to claim 4, wherein the motion vector search device is controlled.