KR100608048B1 - Apparatus for correcting motion vector error and method therefor, and recording medium for recording programs for realizing the same - Google Patents

Abstract

Disclosed are a method and apparatus for correcting a motion vector of a current block using motion information and pattern information of neighboring blocks, and a computer-readable recording medium having recorded thereon a program for implementing the same. The present invention provides a method of correcting a motion vector, comprising: (a) determining whether a current block has a pattern shape; (b) correcting the motion vector of the current block based on the motion vector of the neighboring block of the current block according to the determination in step (a). According to the motion vector correction method according to the present invention, when the error correction of the motion vector of the currently processed block using the motion vector of the neighboring block, by considering the presence or absence of the pattern of the currently processed block, to configure a more natural motion vector field It is possible.

Description

Apparatus for correcting motion vector error and method therefor, and recording medium for recording programs for realizing the same}

1 (a) (b) (c) are diagrams showing a motion estimation error.

FIG. 2 is a block diagram illustrating a motion vector of a current block and a motion vector of a neighboring block to be used as candidate vectors of the current block in the unidirectional motion estimation method.

3 is a diagram for describing a weight generation method used for motion vector error correction.

4 is a block diagram illustrating a motion vector flattening apparatus according to an embodiment of the present invention.

FIG. 5 is a flowchart for describing a motion vector planarization method performed in the motion vector planarizer shown in FIG. 4.

The present invention relates to a method and apparatus for correcting a motion vector, and a recording medium on which a program for implementing the same is recorded. Particularly, a method for adaptively performing motion vector correction in consideration of characteristics of an input image and neighboring blocks of a current block. It is about.

In general, a method used for compressing a video, by comparing the previous video and the current video in units of N × M blocks, estimating how much the current video block is moved from the previous video block, and calculating the coordinates of the current video and the current video. The method of obtaining and transmitting the difference component is used. In this case, since only the difference component is transmitted, the video compression efficiency is greatly increased.

When the motion estimation is properly performed in such a video compression method, the compression efficiency is increased. However, if the motion vector of the estimated current block has a large difference from the motion vector of the neighboring block, the estimated motion vector of the current block is compared with the motion model. It is likely that the area does not match or is an occlusion area in which the direction of the moving object and noise or background parts are staggered. Therefore, in this case, when the estimated motion vector is used as it is, a block artifact occurs in the motion compensator. To remove this, nonlinear filtering such as a median filter may be applied to the estimated motion vector field, but there is still a problem that the motion vector field is not compensated evenly in the occlusion region.

SUMMARY OF THE INVENTION The present invention provides a motion vector error correction method for improving the correction efficiency of a motion vector of each block by adaptively correcting a motion vector of a block in consideration of motion vectors and image characteristics of neighboring blocks. An apparatus and a recording medium having recorded thereon a program for implementing the same are provided.

The technical problem comprises the steps of: (a) determining whether a current block has a pattern form; and (b) correcting the motion vector of the current block based on the motion vector of the neighboring block of the current block according to the determination in step (a).

In addition, the technical problem is an apparatus for correcting a motion vector, comprising: a pattern detector for detecting whether a current block has a pattern shape; And a motion vector correcting unit for correcting the motion vector of the current block based on the motion vector of the neighboring block of the current block according to the detection result of the pattern detecting unit.

In addition, the technical problem is (a) determining whether the current block has a pattern form; (b) correcting the motion vector of the current block based on the motion vector of the neighboring block of the current block according to the determination in step (a). A computer readable recording medium having recorded a program is achieved.

1 (a), (b) and (c) are diagrams showing a motion estimation error caused by influences such as noise and motion occlusion. Fig. 1A is a diagram showing a moving object that is the object of motion vector estimation. 1B is a diagram illustrating an actual motion vector field. 1 (c) is a diagram illustrating an estimated motion vector field.

If the motion estimation is inaccurate in the video compression method using the motion estimation and compensation technique, a motion estimation error as shown in FIG. 1 (c) occurs, which causes block artifacts in motion compensation at the decoding end.

In order to solve this problem, the present invention first considers the smoothness of the motion vector field of the current block and neighboring blocks in motion compensation. That is, assuming that the motion vector of the current block and the motion vector field of the neighboring block are flat, the motion vector of the current block in which the error occurs is replaced with the flat motion vector in consideration of the characteristics of the input image.

2 illustrates a motion vector of a current block and a motion vector of a neighboring block to be used as candidate vectors of the current block in a unidirectional motion estimation method. When the motion between the blocks is flat and the input image does not have a pattern, when there is an error in the motion vector estimation of the current block, the motion vector of the neighboring block with a good estimation may be replaced with the motion vector of the current block.

3 is a view for explaining a weight generation method used for motion vector error correction according to the present invention.

First, in order to obtain a new motion vector of the current processing block 5, the sum of absolute difference (SAD) and the zero vector according to the motion vectors of blocks 1 to 9 and the motion vectors of each block are obtained. That is, the motion vector of the current processing block is calculated and detected.

 The difference value between pixels is calculated by Equation 1 below.

Here, r (m, n) represents data of a reference block (current processing block), and s (m + i, n + j) represents data of a candidate block (block in a search region of a previous image). In this case, the motion vector of the reference block is the motion vector of the block having the smallest SAD value among the motion vectors of all possible candidate blocks in the search region.

The weight used for motion vector error correction is calculated according to Equation 2 below.

Here, denotes a weight of each block, denotes a motion vector of each block, denotes SAD of each block.

According to Equation 2, the weight value for each block is determined according to the SAD value, and there is an inverse relationship. The smaller the SAD value, the better the estimate is, the larger the weight value is. If the SAD value is large and the estimation is incorrect, the weight value is small.

In addition, a weight for determining a new motion vector for the current processing block is calculated by adding weights calculated in Equation 2 to a plurality of candidate vectors, for example, vectors of neighboring blocks. In this embodiment, the new motion vector of the reference block is calculated according to Equation 3 below.

Denotes a new motion vector for the current processing block.

4 is a diagram illustrating a motion vector flattening apparatus 400 according to an embodiment of the present invention.

The motion vector planarization apparatus 400 according to the present invention includes a block weight calculator 420, a pattern detector 440, and a motion vector corrector 460.

The block weight calculator 420 calculates the block weights according to Equation 2 based on the input block motion vector and the SAD information, and outputs the calculated block weights to the motion vector corrector 460.

The pattern detector 440 determines whether the data in the current processing block has the form of a pattern based on the input image data. In this embodiment, the presence or absence of the pattern of neighboring blocks is counted based on the current processing block, and it is determined whether or not the data in the current processing block has the form of the pattern according to the number of blocks having the pattern.

For example, as shown in Fig. 3, when block 5 is the current processing block, 1, 2, 3,. . , Counts how many of the nine blocks of nine have a pattern form, and if the counted number exceeds a predetermined pattern_threshold, the current processing block determines that it has a pattern.

As such, when an error occurs in the motion vector in the grid or the pattern by considering whether the current processing block has a pattern, when the motion vector is corrected using the weight calculated by the block weight calculator 420. The problem that the error correction function is degraded can be prevented from occurring.

The pattern threshold value may be initially set or input by an external user.

The information on the presence or absence of the pattern of the current processing block generated by the pattern detector 440 is input to the motion vector corrector 460.

The motion vector corrector 460 uses the block weight information input by the block weight calculator 420 and information about the presence or absence of a pattern of the current process block input by the pattern detector 440 to determine the motion vector of the current process block. Correct it.

That is, when there is a pattern of the current processing block, the motion vector of the current processing block is output as it is without correction for the motion vector of the current processing block.

If there is a pattern of the current processing block, a new motion vector is generated according to Equation 3, and the generated new motion vector is output as a motion vector of the current processing block.

FIG. 5 is a flowchart for describing a motion vector planarization method performed in the motion vector planarizer shown in FIG. 4.

In operation 520, it is determined whether the data in the current processing block has the form of a pattern based on the input image data. In the present embodiment, the presence or absence of a pattern of neighboring blocks is counted based on the current processing block, and it is determined whether or not data in the current processing block has a pattern form according to the number of blocks having a pattern.

If it is determined that the current processing block has a pattern form, the process proceeds to step 560. If it is determined that the current processing block does not have a pattern form, the process proceeds to step 580.

In operation 540, the weight of each block is calculated according to Equation 2 based on the input motion vector of each block and the SAD information.

In operation 560, a new motion vector is generated according to Equation 3 based on the input weight information for each block, and the generated new motion vector is output.

In operation 580, the motion vector of the current processing block is output as it is without correction for the motion vector of the current processing block.

 The present invention is not limited to the above-described embodiment, and of course, modifications may be made by those skilled in the art within the spirit of the present invention.

The invention can also be embodied as computer readable code on a computer readable recording medium. The computer-readable recording medium includes all kinds of recording devices in which data that can be read by a computer system is stored. Examples of computer-readable recording media include ROM, RAM, CD-ROM, magnetic tape, hard disk, floppy disk, flash memory, optical data storage device, and also carrier waves (for example, transmission over the Internet). It also includes the implementation in the form of. The computer readable recording medium can also be distributed over network coupled computer systems so that the computer readable code is stored and executed in a distributed fashion.

As described above, according to the present invention, when the error correction of the motion vector of the currently processed block using the motion vector of the neighboring block, considering the presence or absence of the pattern of the currently processed block, a flat and more natural without deterioration of the error correction function The effect is that it is possible to construct a motion vector field.

Claims (15)

In the motion vector correction method, (a) determining whether the current block has a pattern form based on the number of neighboring blocks having a pattern form among neighboring blocks adjacent to the current block; (b) if it is determined that the current block does not have a pattern shape, and corrects the motion vector of the current block based on a motion vector of a neighboring block of the current block, and determines that the current block has a pattern shape, And not performing correction on the motion vector of the current block. delete delete delete The method of claim 1, The step (b) includes (b1) calculating a weight based on the motion vector and the SAD value of each of the current block and the neighboring blocks, and (b2) a new motion for the current block based on the calculated weight. Further comprising generating a vector. In the motion vector correction device, A pattern detector for detecting whether the current block has a pattern shape based on the number of neighboring blocks having a pattern shape among neighboring blocks adjacent to the current block; If it is determined that the current block does not have a pattern shape by the pattern detector, the motion vector of the current block is corrected based on the motion vector of a neighboring block of the current block, and the current block has a pattern shape by the pattern detector. And a motion vector correcting unit that does not perform correction on the motion vector of the current block. delete delete delete The method of claim 6, The apparatus may further include a block weight calculator for calculating weights based on the motion vector and the SAD value of each of the current block and neighboring blocks. And the motion vector corrector generates a new motion vector for the current block based on the calculated weight. (a) determining whether the current block has a pattern form based on the number of neighboring blocks having a pattern form among neighboring blocks adjacent to the current block; (b) if it is determined that the current block does not have a pattern shape, and corrects the motion vector of the current block based on a motion vector of a neighboring block of the current block, and determines that the current block has a pattern shape, And a step of not performing a correction on the motion vector of the current block. 17. A computer-readable recording medium having recorded thereon a program for realizing a motion vector correction method. delete delete delete The method of claim 11, The step (b) includes (b1) calculating a weight based on the motion vector and the SAD value of each of the current block and the neighboring blocks, and (b2) a new motion for the current block based on the calculated weight. And further comprising generating a vector.

Images