JPH06311533A - Motion compensation device - Google Patents

Abstract

(57) [Abstract] [PROBLEMS] To provide a motion compensator capable of reducing a difference in motion detection accuracy and coding efficiency caused by a difference between a pixel interval of a luminance signal and a pixel interval of a color difference signal due to a difference in signal format. thing. A motion detection unit setter 8 for setting a motion detection unit, a motion vector detector 7 for detecting a motion vector based on a screen brightness signal and a previous screen brightness signal for each motion detection unit, and a motion vector A color difference motion vector calculator 9 for calculating a color difference motion vector of a color difference signal for each macroblock, and a motion compensation signal for extracting a motion compensation signal from the previous screen for each macroblock based on the color difference motion vector and the motion vector. And take-out devices 10 and 11.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a motion compensator used for high efficiency coding of video signals.

[0002]

2. Description of the Related Art One of video signal coding apparatuses uses motion compensation to reduce redundancy in the time direction.

FIG. 3 is a block diagram of an encoding apparatus for explaining conventional motion compensation. In FIG. 3, 1 is a blocker, 2 is a luminance signal storage unit, 3 is a color difference signal storage unit, 4
Is a frame memory for luminance signals, 5 is a frame memory for color difference signals, 15 is a macroblock setter, 7 is a motion vector detector, 9 is a color difference motion vector calculator, and 10 and 11 are motion compensation for each of the luminance signal and the color difference signals. Signal extractor,
12 is a difference signal calculator, 13 is an encoder, and 14 is a partial decoder.

The operation of the coding apparatus configured as described above will be described.

First, the luminance signal and the color difference signal for one frame to be encoded are rearranged by the blocker 1 and divided into blocks each having 8 pixels vertically and 8 pixels horizontally.

Next, the luminance signal and the color difference signal divided into blocks are stored in the luminance signal storage unit 2 and the color difference signal storage unit 3.
Memorize in each. Here, it is assumed that the luminance signal frame memory 4 and the color difference signal frame memory 5 store the decoded signals of the luminance and the color difference one frame before.

Next, for each of the signals stored in the luminance signal storage unit 2 and the color difference signal storage unit 3, a macroblock setter 15 selects a fixed set of a plurality of luminance signal blocks and color difference signal blocks. It This set is called a macroblock and is a basic unit for motion detection and motion compensation.

In the above-mentioned macro block taking method, for example, a set of a luminance signal block corresponding to the size of one color difference signal block in the screen and a color difference signal block is set as a macro block.

As an example, in the signal format called 4: 2: 2 in which the luminance signal is sampled at 13.5 MHz and the color difference signal is sampled at 6.75 MHz, the ratio of the pixel interval between the luminance signal and the color difference signal is 1: 1 vertically and horizontally. The size of the macroblock is 1: 2, and the size of the macroblock is 8 × 16 pixels in luminance, that is, one luminance signal block × two horizontal luminance signal blocks. As shown in FIG. 4A, the macroblocks in this case include two luminance blocks and chrominance blocks RY and BY.
Each will be a pair. The size of the motion detection unit is 8 × 16 pixels in the luminance signal. Also, FIG.
In this case, (b) shows that the size of the macroblock is 1 vertically.
The size of 6 × 16 pixels indicates that two color difference blocks are vertically gathered.

Further, in the signal format called 4: 1: 1, in which the color difference signal is thinned out to half in comparison with 4: 2: 2, the ratio of the pixel interval between the luminance signal and the color difference signal is 1: 1 in the vertical direction. ,
The width of the macro block is 1: 4, and the size of the macro block is 8 vertical pixels × 32 horizontal pixels in luminance, that is, one luminance signal block × 4 horizontal luminance signal blocks. In this case, the macroblocks are four luminance blocks, as shown in FIG.
One color difference block is provided for each of RY and BY. The unit of motion detection has a size of luminance vertical 8 × horizontal 32 pixels.

Further, in the signal format called 4: 2: 0, in which the color difference signal is thinned out in half in a line-sequential manner compared to 4: 2: 2, the ratio of the pixel interval between the luminance signal and the color difference signal is vertical. The size of the macro block is 1: 2 and the width is 1: 2, and the size of the macro block is 16 pixels in the vertical direction × 16 pixels in the horizontal direction, that is, two luminance signal blocks × two horizontal pixels are collected. As shown in FIG. 4D, the macro blocks in this case are a set of four luminance blocks and one chrominance block for each of RY and BY. The motion detection unit has a size of 16 × 16 pixels of brightness.

Next, with respect to the decoded signal of the luminance of one frame before, which is stored in the luminance signal frame memory 4, the position corresponding to the position of the macroblock to be encoded is used as the reference position, and all of the detection range is detected. The position having the smallest sum of the absolute values of the difference values is detected by the motion vector detector 7 as the motion vector of the luminance.

Subsequently, the color difference motion vector calculator 9 calculates the motion vector of the color difference signal block from the magnitude of the motion vector in the macro block of the luminance signal at the same position detected by the motion vector detector 7.

Here, when the ratio between the pixel interval of the luminance signal and the pixel interval of the color difference signal is 1: 2 in both the vertical and horizontal directions, a vector having a size of 1/2 in both the vertical and horizontal directions corresponds to the motion vector of the luminance. It is defined as the motion vector of the color difference signal block at the position.

Further, when the ratio of the pixel interval of the luminance signal to the pixel interval of the color difference signal is 1: 1 in the vertical direction and 1: 4 in the horizontal direction, the vertical direction is the same and the horizontal direction is 1/4 of the luminance motion vector. The magnitude vector is defined as the motion vector of the color difference signal block at the corresponding position.

If the ratio of the pixel interval of the luminance signal to the pixel interval of the color difference signal is 1: 1 in the vertical direction and 1: 2 in the horizontal direction, it is the same in the vertical direction and 1/2 in the horizontal direction with respect to the motion vector of the luminance. The magnitude vector is defined as the motion vector of the color difference signal block at the corresponding position.

Next, the motion compensation signal extractor 10 uses the position moved from the reference position by the magnitude of the motion vector detected by the motion vector detector 7 as the motion compensation position.
Thus, the motion compensation signal of the luminance signal block having the size of the macro block is extracted from the luminance signal of the reference screen (the screen one before the screen) stored in the luminance signal frame memory 4.

Further, the motion compensation signal extractor 11 causes the color difference signal frame memory 5 to use the position, which has been moved from the reference position by the magnitude of the motion vector detected by the color difference motion vector calculator 9, as the motion compensation position. The motion compensation signal of the color difference signal block is taken out from the color difference signal of the reference screen stored in.

Next, the difference signal calculator 12 calculates the difference between each of the extracted motion compensation signals and the macroblock signals stored in the luminance signal storage unit 2 and the color difference signal storage unit 3. The obtained difference signal is encoded by the encoder 13. Further, the encoded signal is output and is restored to each signal of the original macroblock by the partial decoder 14 and stored in the luminance signal frame memory 4 and the color difference signal frame memory 5. It is used to predict the frame.

[0020]

However, in the above system, if the signal formats handled by the same device are different, the ratio of the number of pixels of the luminance signal and the color difference signal is also different.
There is a problem that there is a large difference between the accuracy of motion compensation and the efficiency of encoding.

For example, when the ratio of the pixel interval between the luminance signal and the color difference signal is 1: 2 in the vertical direction and 1: 2 in the horizontal direction, the size of the luminance signal macroblock is 16 pixels in the vertical direction × 16 pixels in the horizontal direction, and the unit of motion detection is Since it is close to a square, the accuracy of motion detection is considered to be relatively good, but if the ratio of the pixel interval of the luminance signal and the color difference signal is 1: 1 in the vertical direction and 1: 4 in the horizontal direction, the luminance signal macroblock The size is vertical 8 × horizontal 32 pixels,
Since the motion detection unit becomes horizontally long, the difference between the motions at both ends of the macroblock is generally large, and the accuracy of motion detection deteriorates. For this reason, there is a difference in the image quality by motion compensation and the code amount that occurs when encoding is performed using motion compensation.

In consideration of such problems of the conventional motion compensation, the present invention considers such a problem of the conventional motion compensation, and the accuracy of motion detection and the efficiency of encoding caused by the difference between the pixel interval of the luminance signal and the pixel interval of the color difference signal due to the difference of the signal format. It is an object of the present invention to provide a motion compensation device that can reduce the difference between

[0023]

According to the present invention, there is provided a motion compensation area setting means for setting an area having a predetermined size as a basic unit area for motion compensation according to a signal format of an image signal, and the set basic area. A predetermined pixel area equal to or smaller than the size of the unit area,
Motion detection unit setting means for setting as a motion detection area which is a unit of motion detection, and a motion for detecting a motion vector based on a screen brightness signal and a different predetermined screen brightness signal for each of the set motion detection unit areas. Vector detection means, color difference motion vector calculation means for calculating a color difference motion vector of a screen color difference signal for each basic unit area of motion compensation based on the detected motion vector, and the calculated color difference motion vector and motion The motion compensation apparatus includes a motion compensation signal generation unit that generates a motion compensation signal from a different predetermined screen for each basic unit area for motion compensation based on the motion vector detected by the vector detection unit.

[0024]

According to the present invention, the motion compensation area setting means sets a basic unit area for motion compensation in accordance with the signal format of the image signal, and the motion detection unit setting means sets a predetermined unit area equal to or smaller than the size of the basic unit area. The pixel region is set as a motion detection region which is a unit of motion detection, and the motion vector detection means detects a motion vector for each motion detection unit region based on a screen brightness signal and a different predetermined screen brightness signal, The color difference motion vector calculation means calculates the color difference motion vector of the color difference signal of the screen for each basic unit area of motion compensation based on the detected motion vector, and the motion compensation signal generation means calculates the color difference motion vector and the calculated color difference motion vector Based on the detected motion vector, motion compensation signals are generated from different predetermined screens for each basic unit area for motion compensation.

[0025]

DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will be described below with reference to the drawings showing its embodiments.

FIG. 1 is a block diagram showing the configuration of a motion compensation apparatus according to an embodiment of the present invention. In FIG.
1 is a blocker that divides the screen into blocks of 8 × 8 pixels, 2 is a luminance signal storage unit that stores a blocked luminance signal, 3 is a color difference signal storage unit that stores a blocked color difference signal, Reference numeral 4 is a luminance signal frame memory for storing the decoded luminance signal, 5 is a color difference signal frame memory for storing the decoded color difference signal, 6 is a pixel interval of the luminance signal and a pixel interval of the color difference signal, and both pixels A pixel interval determiner for determining the ratio of intervals, 7 a motion vector detector for detecting a motion vector in a luminance signal, 8 a motion detection unit setter for setting the size of a region for detecting a motion vector, 9 a color difference Motion vector in the signal (hereinafter,
A color difference motion vector calculator that calculates a color difference motion vector so that it can be easily distinguished from a motion vector in the luminance signal) is a motion compensation signal for extracting each motion compensation signal based on the motion vector and the color difference motion vector. An extractor, 12 is a difference signal calculator that calculates the difference between the luminance signal and color difference signal, and each motion compensation signal, 13 is an encoder that encodes the obtained difference signal, and 14 is an encoded signal. It is a partial decoder for decoding.

Part of the above-mentioned pixel interval determiner 6 and motion detection unit setting device 8 constitutes motion compensation area setting means, and part of the motion detection unit setting device 8 constitutes motion detection unit setting means. The motion vector detector 7 is the motion vector detecting means, the color difference motion vector calculator 9 is the color difference motion vector calculating means, the motion compensation signal extractors 10 and 11 constitute the motion compensation signal generating means, and the pixel interval is The determiner 6 constitutes a pixel interval input means.

Next, the operation of the motion compensation device of the above embodiment will be described.

In this case, the motion compensating apparatus uses as an input signal a signal format in which the ratio of the pixel interval of the luminance signal to the pixel interval of the color difference signal is 1: 2 both in the vertical and horizontal directions, such as 4: 2: 0.
The signal format in which the ratio of the pixel interval of the luminance signal to the pixel interval of the color difference signal is 1: 1 in the vertical direction and 1: 4 in the horizontal direction such as 1: 1 and 4:
It is assumed that three types of signals in which the ratio of the pixel interval of the luminance signal and the pixel interval of the color difference signal is 1: 1, vertically and 1: 2 horizontally are switched and used as in 2: 2 (see FIG. a),
(See (c) and (d)).

First, the luminance signal and the color difference signal for one frame to be encoded are rearranged by the blocker 1 and divided into blocks each having 8 pixels vertically and 8 pixels horizontally.

Next, the luminance signal and the color difference signal divided into blocks are stored in the luminance signal storage unit 2 and the color difference signal storage unit 3, respectively.

At this time, it is assumed that the luminance signal frame memory 4 and the color difference signal frame memory 5 store the decoded signals of the luminance and the color difference one frame before.

On the other hand, the pixel interval determiner 6 is inputted with switching information of the signal format to be handled from the outside by the user,
Thereby, the pixel interval of the luminance signal and the pixel interval of the color difference signal in the screen, and the ratio of the pixel intervals of the both are determined. The size of the macroblock is determined based on this determination result.

Next, with respect to the luminance signal of the reference screen (here, the previous screen) stored in the luminance signal frame memory 4, the position corresponding to the position of the block to be encoded is used as the standard position. The motion vector detector 7 detects a position having the smallest sum of absolute difference values among all positions within the detection range as a motion vector of the luminance signal.

At this time, the motion detection unit setting unit 8 sets a size in which two blocks of vertical 8 × horizontal 8 pixels of the luminance signal are arranged side by side as a motion detection unit (see a portion indicated by a dotted line in FIG. 2). ).

Next, the color difference motion vector calculator 9
The color difference motion vector of the color difference signal block is calculated from the size of the motion vector in the macro block of the luminance signal at the same position.

Here, as shown in FIG. 2D, when the ratio between the pixel interval of the luminance signal and the pixel interval of the color difference signals is 1: 2 in both the vertical and horizontal directions, the size of the macroblock is set to the vertical direction of the luminance signal. × 16 horizontal pixels, 8 color difference signals × 8 horizontal pixels.
This size is equivalent to four blocks each consisting of two blocks vertically and horizontally, that is, two vertical motion detection units.The average value of the two vertical motion detection units is calculated, and the obtained average value is calculated. A vector with a size of 1/2 in both the vertical and horizontal directions,
It is defined as the color difference motion vector of the color difference signal block at the corresponding position.

Further, as shown in FIG. 2C, when the ratio between the pixel interval of the luminance signal and the pixel interval of the color difference signal is 1: 1 in the vertical direction and 1: 4 in the horizontal direction, the size of the macro block is set. The luminance signal has vertical 8 × horizontal 32 pixels, and the color difference signal vertical 8 × horizontal 8 pixels. This size is a collection of four luminance signal blocks horizontally,
That is, the average value of the two motion detection unit horizontal motion vectors is calculated, and the average value of the two horizontal motion detection units is calculated. It is defined as the color difference motion vector of the color difference signal block at the corresponding position.

Further, as shown in FIG. 2A, when the ratio between the pixel interval of the luminance signal and the pixel interval of the color difference signal is 1: 1 in the vertical direction and 1: 2 in the horizontal direction, the size of the macro block is set. The luminance signal has vertical 8 × horizontal 16 pixels, and the color difference signal vertical 8 × horizontal 8 pixels. This size is a collection of two luminance signal blocks,
That is, a vector corresponding to one motion detection unit and having the same vertical length and 1/2 horizontal size as the motion vector of the motion detection unit is determined as the color difference motion vector of the color difference signal block at the corresponding position.

Next, from the luminance signal of the reference screen stored in the luminance signal frame memory 4, the position moved from the reference position by the magnitude of the motion vector detected by the motion vector detector 7 is used as the motion compensation position, The motion compensation signal extractor 10 extracts the motion compensation signal of the luminance signal block having the size of the macro block.

From the color difference signal of the reference screen stored in the color difference signal frame memory 5, the position moved from the reference position by the size of the color difference motion vector calculated by the color difference motion vector calculator 9 is used as the motion compensation position. , The motion compensation signal of the color difference signal block is extracted by the motion compensation signal extractor 11.

Thereafter, the difference signal calculator 12 divides the motion compensation signals of the respective signals extracted as described above and the macroblock signals stored in the luminance signal storage unit 2 and the color difference signal storage unit 3. Each difference is calculated. The difference signal thus obtained is encoded by the encoder 13.

The coded signal is output as a coded signal and is restored to the original macroblock signal by the partial decoder 14, and the decoded signal of the luminance and the color difference is used as the luminance signal frame memory 4. And the color difference signal frame memory 5 are used to predict the next frame.

As described above, even when different signal formats are handled on the same device, the motion detection of the luminance signal should be performed with a common size in which two blocks each having 8 × 8 pixels are arranged horizontally. Since the motion detection unit is the same regardless of the signal format, the difference in motion detection accuracy is small even if the signal format is different. Further, only the average value needs to be calculated when calculating the color difference motion vector of the color difference signal, and the increase in the device configuration is small and the circuit is not complicated.

In the above embodiment, the size of the divided block is 8 × 8 pixels, but the size is not limited to this.

In the above embodiment, the signal format switching information is input from the outside by the user. Instead of this, the size of the currently used image, the pixel interval, etc. are measured. A configuration may be provided in which a determination unit is provided and the signal format switching information is automatically input.

Further, in the above embodiment, the sum of absolute values of the difference values is compared with respect to all the positions within the detection range. However, the present invention is not limited to this, and weighting is performed with the reference position as the center, or a number is added. A step tree search method may be used.

In the above embodiment, the sum of absolute values of difference values is compared in motion vector detection and the smallest one is selected. However, the present invention is not limited to this, and other functions such as the sum of squares of difference values may be used. You may use. Alternatively, as the comparison method, a method of preparing one or a plurality of threshold values and selecting one from the respective judgments may be used.

In the above embodiment, the ratio of the pixel interval of the luminance signal and the pixel interval of the color difference signal is 1: 1 vertically and 1: 2 horizontally.
In the case of, the size of the macroblock is 8 ×
Assuming that 16 pixels in the horizontal direction and 8 pixels in the vertical direction of the color difference signal × 8 pixels in the horizontal direction are defined as the motion vector of the color difference signal block at the corresponding position, a vector having the same length in the vertical direction and a size of 1/2 in the horizontal direction is defined. However, instead of this, FIG.
As shown in,
The average value of the two motion detection unit vertical motion vectors is calculated as 6 × horizontal 16 pixels, color difference signal vertical 16 × horizontal 8 pixels, and the vertical is the same and the horizontal is 1/2 the size of the obtained average. May be defined as a motion vector for two vertical blocks of the color difference signal at the corresponding position.

Further, in the above-mentioned embodiment, each means is constituted by dedicated hardware, but instead of this, similar functions may be realized by software using a computer.

[0051]

As is apparent from the above description, according to the present invention, it is possible to reduce the difference in motion detection accuracy and coding efficiency caused by the difference in pixel interval between the luminance signal and the color difference signal due to the difference in signal format. It has advantages.

[Brief description of drawings]

FIG. 1 is a block diagram showing the configuration of a motion compensation device according to an embodiment of the present invention.

2 (a), (b), (c), and (d) are diagrams for explaining pixel arrangement and motion detection unit in each signal format in the same embodiment.

FIG. 3 is a block diagram showing a configuration of a conventional motion compensation device.

4 (a), (b), (c) and (d) are diagrams for explaining pixel arrangement and motion detection unit in each signal format in a conventional motion compensation device.

[Explanation of symbols]

 1 Blocking device 2 Luminance signal storage unit 3 Color difference signal storage unit 4 Luminance signal frame memory 5 Color difference signal frame memory 6 Pixel interval determination unit 7 Motion vector detector 8 Motion detection unit setting unit 9 Color difference motion vector calculator 10, 11 Motion compensation signal extractor

Claims (4)

[Claims] 1. A motion compensation area setting means for setting an area having a predetermined size as a basic unit area for motion compensation according to a signal format of an image signal, and a predetermined value equal to or smaller than the size of the set basic unit area. Motion detection unit setting means for setting the pixel region of the above as a motion detection region which is a unit of motion detection, and for each motion detection unit region that has been set, a motion is detected based on a screen brightness signal and a different predetermined screen brightness signal. Motion vector detecting means for detecting a vector, color difference motion vector calculating means for calculating a color difference motion vector of the color difference signal of the screen for each basic unit area of the motion compensation based on the detected motion vector, and its calculation Based on the color difference motion vector and the motion vector detected by the motion vector detection means, for each basic unit area of the motion compensation,
A motion compensation apparatus comprising: a motion compensation signal generation unit that generates a motion compensation signal from the different predetermined screen. 2. The motion compensation area setting means has a pixel interval input means for inputting a pixel interval of a luminance signal and a pixel interval of a color difference signal in the signal format, and a ratio of the pixel intervals. Item 1. The motion compensation device according to item 1. 3. The motion detection area has a predetermined natural number m, n.
2. The motion compensating apparatus according to claim 1, wherein two regions formed by m × n pixels are arranged. 4. A pixel interval ratio exists in both the vertical and horizontal directions of the screen, and the color difference motion vector calculation means detects the motion when the pixel interval ratio of the luminance signal and the color difference signal is 1: 2 in both the vertical and horizontal directions. A color difference motion vector at a corresponding position is calculated based on an average value of two vertically aligned motion vectors in the unit area, and the ratio of the pixel interval between the luminance signal and the color difference signal is 1 in the vertical direction. In the case of 1, and the width is 1: 2, the color difference motion vector at the corresponding position is calculated based on the motion vector of the motion detection unit area, and the pixel interval ratio of the luminance signal and the color difference signal is 1: 1 in the vertical direction. If the width is 1: 4,
3. The motion compensation apparatus according to claim 2, wherein the color difference motion vector at the corresponding position is calculated based on the average value of two motion vectors that are arranged in a row in the motion detection unit.