Video signal decoding method and device

Patent metadata

Field	Value
Publication number	US-9544593-B2
Application number	US-201013517313-A
Country	US
Kind code	B2
Filing date	Dec 15, 2010
Priority date	Dec 20, 2009
Publication date	Jan 10, 2017
Grant date	Jan 10, 2017

Abstract

According to the present invention, when the block merge flag of the current prediction block specifies that the motion information of the current prediction block is acquired using block merging, the motion information of the current prediction block is acquired from the motion information of the neighboring block, a reference block of a reference picture is acquired using the motion information of the current prediction block, the prediction value of the current prediction block is acquired using the reference block, and the current prediction block is restored using the residual of the current prediction block and the prediction value of the current prediction block.

First claim

The invention claimed is:

1. A method of decoding a video signal, the method comprising:
obtaining motion information of a current prediction block of a current picture;
obtaining at least two reference blocks from reference pictures of the current prediction block using the motion information of the current prediction block;
obtaining first weight information using a difference between the current prediction block and a first reference block of the at least two reference blocks;
obtaining a first offset value for the first reference block from a difference between an average of pixel values of a template region of the first reference block and an average of pixel values of a template region of the current prediction block, the template region of the first reference block being a region in at least one neighboring block adjacent to the first reference block, and the template region of the current prediction block being a region in at least one neighboring block adjacent to the current prediction block;
obtaining second weight information using a difference between the current prediction block and a second reference block of the at least two reference blocks;
obtaining a second offset value for the second reference block from a difference between an average of pixel values of a template region of the second reference block and the average of pixel values of the template region of the current prediction block, the template region of the second reference block being a region in at least one neighboring block adjacent to the second reference block;
obtaining a prediction value of the current prediction block by applying the obtained first weight information to pixel values of the first reference block and the first offset value and applying the obtained second weight information to pixel values of the second reference block and the second offset value, wherein the prediction value of the current block is obtained from a sum of the weighted pixel values of the first reference block, the weighted first offset value, the weighted pixel values of the second reference block, and the weighted second offset value; and
reconstructing the current prediction block using the prediction value of the current prediction block and a residual data of the current prediction block.

2. The method of claim 1 , wherein the obtaining of the motion information of the current prediction block comprises:
acquiring a block merge flag of the current prediction block; and
acquiring the motion information of the current prediction block using motion information of a neighboring block when the block merge flag indicates that the motion information of the current prediction block is acquired using block merging,
wherein the block merge flag corresponds to information that indicates whether the motion information of the current prediction block is derived using block merging, the block merging means merging of the current prediction block and the neighboring block, and the neighboring block includes a left neighboring block and an upper neighboring block of the current prediction block.

3. The method of claim 2 , wherein the obtaining of the motion information of the current prediction block comprises:
checking whether motion information of the left neighboring block is equal to motion information of the upper neighboring block; and
acquiring a left block merge flag of the current prediction block when the motion information of the left neighboring block is not equal to the motion information of the upper neighboring block,
wherein the left block merge flag indicates whether the current prediction block is merged with the left neighboring block or the upper neighboring block.

4. The method of claim 3 , wherein the motion information of the current prediction block is obtained from the motion information of the left neighboring block when the left block merge flag indicates that the current prediction block is merged with the left neighboring block, and the motion information of the current prediction block is obtained from the motion information of the upper neighboring block when the left block merge flag indicates that the current prediction block is merged with the upper neighboring block.

5. The method of claim 1 , wherein the motion information includes at least two motion vectors and at least two reference picture indices.

6. The method of claim 5 , wherein when a motion vector having decimal pixel resolution corresponding to one of ½ pixel resolution, ¼ pixel resolution, or ⅛ pixel resolution, the obtaining of the at least two reference blocks from the reference pictures comprises generating a signal having decimal pixel resolution from a corresponding reference picture using an interpolation filter,
wherein the interpolation filter multiplies an input signal thereof by filter coefficients for every pixel and outputs the sum of multiplication results.

7. The method of claim 6 , wherein the input signal of the interpolation filter includes at least one of integer pixels of the corresponding reference picture and a previously generated signal having decimal pixel resolution.

8. The method of claim 6 , wherein the interpolation filter is used in the vertical direction, horizontal direction or diagonal direction in consideration of position information on the signal having decimal pixel resolution,
wherein the position information on the signal having decimal pixel resolution corresponds to a difference between the position of an integer pixel and the position of the signal having decimal pixel resolution.

9. The method of claim 6 , wherein the interpolation filter is specified by interpolation filter identification information,
wherein the interpolation filter identification information indicates a type or filter coefficient of the interpolation filter.

10. The method of claim 1 , wherein a slice type of the current prediction block corresponds to a bidirectional prediction type slice.

11. The method of claim 1 , wherein the difference between the current prediction block and the first reference block includes a difference between a sum of the pixel values of the current prediction block and a sum of the pixel values of the first reference block.

12. The method of claim 1 , wherein the difference between the current prediction block and the second reference block includes a difference between a sum of the pixel values of the current prediction block and a sum of the pixel values of the second reference block.

13. The method of claim 1 , wherein the difference between the current prediction block and the first reference block includes a difference between an average of the pixel values of the current prediction block and an average of the pixel values of the first reference block.

14. The method of claim 1 , wherein the difference between the current prediction block and the second reference block includes a difference between an average of the pixel values of the current prediction block and an average of the pixel values of the second reference block.

15. The method of claim 1 , wherein the first weight information is obtained using the following equation:
W 0= D 1/( D 0+ D 1),

where W 0 denotes the first weight information, D 0 denotes the difference between the current prediction block and the first reference block, and D 1 denotes the difference between the current prediction block and the second reference block.

16. The method of claim 1 , wherein the second weight information is obtained using the following equation:
W 1= D 0/( D 0+ D 1),
where W 1 denotes the second weight information,

Assignees

• Park Seung Wook
• Lim Jae Hyun
• Kim Jung Sun
• Park Joon Young
• Choi Young Hee
• Sung Jae Won
• Jeon Byeong Moon
• Jeon Yong Joon
• Lg Electronics Inc

Inventors

• Park Seung Wook
• Lim Jae Hyun
• Kim Jung Sun
• Park Joon Young
• Choi Young Hee
• Sung Jae Won
• Jeon Byeong Moon
• Jeon Yong Joon

Classifications

• H04N19/573

  Motion compensation with multiple frame prediction using two or more reference frames in a given prediction direction · CPC title

• H04N19/52

  by predictive encoding · CPC title

• H04N19/46

  Embedding additional information in the video signal during the compression process (H04N19/517, H04N19/68, H04N19/70 take precedence) · CPC title

• H04N19/577

  Motion compensation with bidirectional frame interpolation, i.e. using B-pictures · CPC title

• H04N19/523

  with sub-pixel accuracy · CPC title