JPH05236452A - Method and device for detection motion vector - Google Patents

Abstract

PURPOSE:To efficiently detect a motion vector with high accuracy. CONSTITUTION:A motion vector detection circuit 52 detects the motion vector between a current frame and a reference frame in continuous frames according to a block matching method, and a judging circuit 54 judges the evaluation function value of this detected motion vector in respect to a threshold value. Based on the judge signal of this judging circuit 54, a control circuit 56 controls whether motion vector detection with much higher accuracy is executed by the above-mentioned motion vector detection circuit 52 or not.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method and a device for detecting a motion vector of digital image data, and more particularly to a method and a device for detecting a motion vector with high accuracy.

[0002]

2. Description of the Related Art As a moving picture signal coding method, a motion-compensated interframe coding method has been actively researched and its effectiveness is known.

This motion-compensated interframe coding system utilizes the correlation between frames (or fields) of a moving image to calculate the amount of motion between a current frame and a reference frame located in the past or in the future. The present frame is compensated based on this motion amount, the difference between this frame and the current frame is calculated, and this difference value and motion information are encoded.

In order to perform this motion compensation, it is important to obtain the motion amount (moving vector) of the moving image signal with high accuracy, and a method called a block matching method for obtaining the moving vector in block units is well known. There is.

FIGS. 4A and 4B show a principle diagram of this method. That is, this method corresponds to all the pixels 110 within a certain search range 108 centered on the pixel 106 at the same position of the reference frame 104 with respect to the block 102 whose motion vector is to be detected in the current frame 100. With respect to the candidate vector 112, a block-based evaluation function value is calculated based on the difference between frames, and the candidate vector having the minimum evaluation function value is detected as the motion vector 114 of the block. Generally, since the matching calculation is performed on a pixel-by-pixel basis, the amount of motion that can be detected is also accurate to one pixel. However, when the movement is complicated, 1
Pixel-accurate motion vectors are not enough, and higher-precision motion vectors are required.

Therefore, a motion vector detecting method with 1/2 pixel accuracy has been proposed. This method, as shown in FIG.
First, block matching is performed at the position of the integer pixel 116 to detect the first motion vector 118. The motion vector 118 has one pixel precision. Next, with respect to the first motion vector 118, eight motion vectors 120 separated by 1/2 pixel are evaluated. Then, of the first motion vector 118 and the eight candidate motion vectors with 1/2 pixel accuracy, the one having the smallest evaluation value is set as the finally detected motion vector 122.

[0007]

By the way, in order to detect a half-pixel precision motion vector as described above, a half-pixel value required for block matching is generated from a plurality of integer pixel values. This must be done, but this process is complicated, and the processing time cannot be denied. Moreover, in the conventional method of detecting a motion vector with 1/2 pixel accuracy, it is meaningless to increase the detection accuracy of a stationary block, for example, because all blocks are subjected to uniform processing. Even for blocks that have a sufficiently accurate motion vector with 1 pixel accuracy
There is a problem in that wasteful processing is performed because detection is performed with a precision of / 2 pixels.

The present invention has been made in view of the above points, and an object of the present invention is to provide a motion vector detecting method and a motion vector detecting apparatus capable of efficiently detecting a motion vector with high accuracy.

[0009]

In order to achieve the above object, in the motion vector detecting method of the present invention, the current attention of the digital image data consisting of a plurality of frames which are sequentially input in time series. Regarding a frame, when the motion vector indicating the motion of each block is detected by the block matching method with reference to the past or future frame, the motion vector detection is performed for each block with one pixel accuracy to obtain the first motion vector. With the accuracy of less than 1 pixel in the surroundings, centering around the obtained step and the obtained first motion vector,
Further, a step of determining whether or not to detect a motion vector, and, if it is determined to detect with higher accuracy, evaluate the motion vector with an accuracy of 1 pixel or less around the first motion vector. And detecting a highly accurate motion vector.

Further, the motion vector detecting apparatus according to the present invention comprises a motion vector detecting circuit for detecting a motion vector between a current frame in continuous frames and a reference frame by a block matching method, and the motion vector detecting circuit. A determination circuit that determines the evaluation function value of the detected motion vector with respect to a threshold value, and whether or not to perform more accurate motion vector detection by the motion vector detection circuit based on the determination signal of the determination circuit. And a control circuit for controlling.

[0011]

In the motion vector detecting method and the motion detecting device according to the present invention, after the motion vector is detected by the block matching with one pixel accuracy, it is determined for each block whether or not the detection is performed with higher accuracy. Therefore, it is possible to efficiently perform highly accurate motion vector detection without wasteful processing.

[0012]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings.

FIG. 2 is a diagram for explaining a pixel arrangement during block matching according to an embodiment of the present invention. In the figure, the current block 10 is the block in the current frame for which the motion vector is being obtained. Reference numeral 12 is a pixel in the reference frame, which is a pixel at the center of the block obtained when motion vector detection is performed with one pixel accuracy, and 14 is a first motion vector. 16 is a 1/2 pixel which is 1/2 pixel away from the detection pixel 12 and 18 is 1
It is a candidate motion vector of / 2 pixel accuracy. Reference numeral 20 is an example of a reference block set centering on the 1/2 pixel 16 and 22 is a 1/2 pixel forming the reference block 20.

FIG. 3 is a diagram for explaining how to calculate the 1/2 pixel value. The integer pixels 30 to 36 are input pixel data and are located at integer positions. Also, 1/2
The pixels 38 to 42 are located at a distance of 1/2 pixel from the surrounding integer pixels, and are generated by taking the average value of the surrounding integer pixels.

FIG. 1A is a flow chart of processing in motion vector detection. Below, using this figure,
The processing procedure of this embodiment will be described. First, the input image is divided into a plurality of blocks, and block matching with 1-pixel accuracy is performed for each block to detect the first motion vector that minimizes the evaluation function value (step S1).

Next, it is determined whether the minimum evaluation function value is "0" or less than a preset threshold value (step S
2) If the evaluation function value is "0" or less than the threshold value, it is determined that it is not necessary to further improve the detection accuracy,
The motion vector detection process ends.

On the other hand, if it is determined in step S2 that the evaluation function value is greater than or equal to the threshold value, it is considered that detection with one pixel accuracy is insufficient, and motion vector detection with 1/2 pixel accuracy is performed. Is performed (step S3).

In this 1 / 2-pixel precision detection, as shown in FIG. 2, the evaluation function value is calculated for eight vectors 18 that are 1/2 pixel apart around the first motion vector 14, and
The motion vector having the smallest evaluation function value is detected from among the nine vectors including the motion vector 14 of. The 1/2 pixel value used at this time is generated by averaging two or four adjacent integer pixel values, as shown in FIG. Further, as the evaluation function, the sum of absolute differences, the sum of squared differences, and the like can be considered.

As described above, after detecting the motion vector with the one-pixel precision (first motion vector), the evaluation function value is evaluated to determine whether or not the detection with the half-pixel precision is further performed. High-precision motion vector detection can be efficiently performed without performing unnecessary processing for motion vectors that have been sufficiently detected with one pixel accuracy.

In the above embodiment, the case where the precision of the high precision motion vector detection is set to 1/2 pixel is illustrated.
Needless to say, 4-pixel precision or the like may be used. Also, 1/2
It is also conceivable to further improve the accuracy in multiple stages, such as performing further determination after detecting with pixel accuracy and detecting with 1/4 pixel accuracy.

Next, a second embodiment of the present invention will be described.
FIG. 1B is a block configuration diagram of a motion vector detection device to which the present invention is applied. In the figure, 50 is an image memory for storing a plurality of frames of moving image data, and 52 is an image memory.
Is a motion vector detection circuit. Further, 54 is a determination circuit, and 56 is a control circuit. Here, the determination circuit 5
4 determines the evaluation function value obtained by the motion vector detection circuit 52 with respect to a preset threshold value. The control circuit 56 controls each component.

The operation of the above configuration will be described below. At the same time as the detection is started, the control circuit 56 works so that the image memory 5
The values of the current frame stored in 0 and the past frame referred to are read to the motion vector detection circuit 52 for each block. Here, the value of the past frame is obtained by passing the value stored in the image memory 50 through the delay circuit 58.

In the motion vector detection circuit 52, first, block matching is performed on a pixel-by-pixel basis to calculate an evaluation function value to obtain a motion vector (first motion vector). Also,
The obtained evaluation function value is sent to the determination circuit 54. The determination circuit 54 compares the given evaluation function value with a threshold value and outputs a determination signal to the control circuit 56 as to whether the evaluation function value is equal to or less than the threshold value. If the determination signal is true, the control circuit 56 terminates the detection process for the block and controls the image memory 50 and the motion vector detection circuit 52 to move to the process of the next block. At the same time, the detected motion vector (first motion vector) is output.

On the other hand, if the determination signal given from the determination circuit 54 is false, the control circuit 56 further commands the motion vector detection circuit 52 to perform detection with 1/2 pixel accuracy. As a result, the motion vector detection circuit 52 generates a 1/2 pixel value, performs block matching with 1/2 pixel accuracy, and obtains and outputs an optimum motion vector. The motion vector detection process is performed by sequentially repeating the above process for each block.

[0025]

As described in detail above, according to the present invention,
It is possible to provide a motion vector detection method and a detection device capable of efficiently detecting a motion vector with high accuracy. That is, after detecting the motion vector with one pixel accuracy, it is determined whether or not the detection with higher accuracy is performed for each block. Vector detection can be performed.

Therefore, the detection processing time can be shortened as compared with the case where highly accurate detection is performed for all blocks. In addition, the structure of the motion vector detecting device can be simplified and power consumption can be reduced.

[Brief description of drawings]

FIG. 1A is a flowchart of a motion vector detection method of the present invention, and FIG. 1B is a block configuration diagram of a motion vector detection device of the present invention.

FIG. 2 is a diagram illustrating a pixel arrangement at the time of block matching according to an embodiment of the present invention.

FIG. 3 is a diagram illustrating a method of calculating a 1/2 pixel value.

FIG. 4 is a diagram for explaining a conventional block matching method, FIG. 4A is a diagram showing a relationship between frames and blocks;
(B) is a figure for demonstrating a candidate vector.

FIG. 5 is a diagram for explaining a conventional 1 / 2-pixel precision motion vector detection method.

[Explanation of symbols]

50 ... Image memory, 52 ... Motion vector detection circuit, 54 ...
Judgment circuit, 56 ... Control circuit.

Claims (2)

[Claims] 1. A motion vector indicating a motion of each block by a block matching method with respect to a current frame of interest of digital image data composed of a plurality of frames sequentially input in time series is referred to by a past or future frame. At the time of detection, a step of performing a motion vector detection for each block with a pixel accuracy to obtain a first motion vector, and a step with a precision of 1 pixel or less around the periphery of the first motion vector obtained, And a step of determining whether or not to detect a motion vector, and if it is determined to detect the motion vector with higher accuracy, evaluate the motion vector with accuracy of 1 pixel or less around the first motion vector. And a step of detecting a highly accurate motion vector, the motion vector detecting method comprising: 2. A current frame among consecutive frames,
A motion vector detection circuit that detects a motion vector between reference frames by a block matching method, a determination circuit that determines the evaluation function value of the motion vector detected by the motion vector detection circuit with respect to a threshold value, and the determination circuit A motion vector detection device, comprising: a control circuit that controls whether or not the motion vector detection circuit performs more accurate motion vector detection based on a determination signal.