Image encoding method and apparatus and image decoding method and apparatus based on motion vector normalization

• US 9,571,832 B2
• Filed: 07/09/2014
• Issued: 02/14/2017
• Est. Priority Date: 01/09/2012
• Status: Active Grant

• Current Assignee: Korea Advanced Institute of Science and Technology, Samsung Electronics Co. Ltd. (Samsung Group)
• Original Assignee: Korea Advanced Institute of Science and Technology, Samsung Electronics Co. Ltd. (Samsung Group)
• Inventors: Park, Seung-ho, Lim, Hyung-jun, Kim, Jae-hyun, Park, Se-hyeok, Park, Hyun-wook
• Primary Examiner(s): Williams, Jeffery
• Application Number:
  US14/327,074
• Publication Number:
  US 20140321548A1
• Time in Patent Office: 951 Days
• Field of Search: 375/240.16
• US Class Current: 1/1
• CPC Class Codes: H04N 19/119 Adaptive subdivision aspects, e.g. subdivision of a picture into rectangular or non-rectangular coding blocks
• CPC Class Codes: H04N 19/139 Analysis of motion vectors, e.g. their magnitude, direction, variance or reliability
• CPC Class Codes: H04N 19/147 according to rate distortion criteria rate-distortion as a criterion for motion estimation H04N19/567
• CPC Class Codes: H04N 19/176 the region being a block, e.g. a macroblock
• CPC Class Codes: H04N 19/192 the adaptation method, adaptation tool or adaptation type being iterative or recursive
• CPC Class Codes: H04N 19/513 Processing of motion vectors
• CPC Class Codes: H04N 19/52 by predictive encoding
• CPC Class Codes: H04N 19/521 for estimating the reliability of the determined motion vectors or motion vector field, e.g. for smoothing the motion vector field or for correcting motion vectors
• CPC Class Codes: H04N 19/543 using regions

Abstract

Provided are an image encoding method and apparatus, and an image decoding method and apparatus. The image encoding method includes: obtaining a variation of a motion estimation error with respect to a homogeneity degree of a motion vector according to mergence, based on a motion vector of each of merged blocks obtained by merging adjacent blocks, motion vectors of the adjacent blocks included in each of the merged blocks, and motion vectors of neighboring blocks of each of the merged blocks; and determining, based on the obtained variation, a merged block having a smallest variation from among the merged blocks and whether to merge the adjacent blocks included in the determined merged block.

Claims

1. An image encoding method comprising:

obtaining motion vectors of blocks forming an image;

obtaining merged blocks by merging adjacent blocks;

obtaining a motion vector of each of the merged blocks;

obtaining a variation of a motion estimation error with respect to a homogeneity degree of motion vectors according to mergence with respect to each of the merged blocks, based on the motion vector of each of the merged blocks, motion vectors of the adjacent blocks included in each of the merged blocks, motion vectors of neighboring blocks of each of the merged blocks, and motion vectors of neighboring blocks of the adjacent blocks included in each of the merged blocks; and

determining a merged block having a smallest variation from among the merged blocks, and determining whether to merge blocks included in the determined merged block, based on the obtained variation.

2. The image encoding method of claim 1, wherein the obtaining of the variation comprises:

obtaining a homogeneity degree of a motion vector by using the motion vectors of the adjacent blocks included in each of the merged blocks, the motion vectors of the neighboring blocks of each of the merged blocks, and the motion vectors of the neighboring blocks of the adjacent blocks included in each of the merged blocks, wherein the homogeneity degree indicates a difference between a degree of non-homogeneity between the motion vector of each of the merged blocks and the motion vectors of the neighboring blocks of each of the merged blocks, and a degree of non-homogeneity between each of the motion vectors of the adjacent blocks included in each of the merged blocks and each of the motion vectors of the neighboring blocks of the adjacent blocks included in each of the merged blocks;

obtaining a variation of a motion estimation error, which indicates a difference between a motion estimation error by the motion vector of each of the merged blocks and a motion estimation error by each of the motion vectors of the adjacent blocks included in each of the merged blocks; and

obtaining the variation of the motion estimation error with respect to the homogeneity degree of the motion vector according to the mergence, by dividing the obtained variation by the obtained homogeneity degree.

3. The image encoding method of claim 2, wherein, when T denotes a merged block from among each of the merged blocks, R and S respectively denote two blocks included in the merged block T, V_T denotes the motion vector of the merged block T, V_R and V_S respectively denote motion vectors of the two blocks R and S, SAD(T, V_T) denotes a motion estimation error by the motion vector V_T of the merged block T, SAD(R, V_R) denotes a motion estimation error by the motion vector V_R of the block R, SAD(S, V_S) denotes a motion estimation error by the motion vector V_S of the block S, φ

(X) denotes a function for obtaining a degree of non-homogeneity of a motion vector of a block X and a motion vector of a neighboring block of the block X, which is obtained according to an equation

4. The image encoding method of claim 1, wherein the determining comprises comparing the obtained variation with a predetermined threshold value, and merging the blocks included in the determined merged block when the obtained variation is smaller than the predetermined threshold value, and maintaining an original block without merging the blocks included in the determined merged block when the obtained variation is equal to or higher than the predetermined threshold value.

5. The image encoding method of claim 1, wherein operations of the image encoding method are repeatedly performed while the obtained variation is lower than the predetermined threshold value.

6. The image encoding method of claim 1, further comprising encoding mergence information of the blocks and motion vector information of the merged blocks, in a bitstream obtained by encoding the image.

13. A non-transitory computer-readable recording medium having recorded thereon a program for executing a method of claim 1.

7. An image encoding apparatus comprising:

a motion estimator and compensator that is configured to obtain motion vectors of blocks forming an image, and a motion vector of each of merged blocks obtained by merging adjacent blocks; and

a block merger that is configured to obtain a variation of a motion estimation error with respect to a homogeneity degree of a motion vector according to mergence with respect to each of the merged blocks, based on the motion vector of each of the merged blocks, motion vectors of the adjacent blocks included in each of the merged blocks, motion vectors of neighboring blocks of each of the merged blocks, and motion vectors of neighboring blocks of the adjacent blocks included in each of the merged blocks, and determine, based on the obtained variation, a merged block having a smallest variation from among the merged blocks and whether to merge blocks included in the determined merged block.

8. An image decoding method comprising:

obtaining information about a merged block obtained by merging adjacent blocks from among blocks forming an image and about a motion vector of the merged block, from a received bitstream; and

obtaining an estimation value of the merged block by performing motion compensation on the merged block by using the motion vector,

wherein the merged block is obtained by merging the adjacent blocks based on a result of comparing a variation of a motion estimation error with respect to a homogeneity degree of a motion vector according to the merging of the adjacent blocks with a predetermined threshold value.

9. The image decoding method of claim 8, wherein the variation is obtained by dividing a variation of a motion estimation error indicating a difference between a motion estimation error by the motion vector of the merged block and a motion estimation error by a motion vector of each of the adjacent blocks included in the merged block by a homogeneity degree of a motion vector indicating a difference between a degree of non-homogeneity between the motion vector of the merged block and motion vectors of neighboring blocks of the merged block, and a homogeneity degree between the motion vector of each of the adjacent blocks included in the merged block and a motion vector of each of neighboring blocks of the blocks forming the image.

10. The image decoding method of claim 9, wherein, when T denotes the merged block, R and S respectively denote two blocks included in the merged block, V_T denotes the motion vector of the merged block, V_R and V_S respectively denote motion vectors of the two blocks R and S, SAD(T, V_T) denotes a motion estimation error by the motion vector V_T of the merged block T, SAD(R, V_R) denotes a motion estimation error by the motion vector V_R of the block R, SAD(S, V_S) denotes a motion estimation error by the motion vector V_S of the block S, φ

(X) denotes a function for obtaining a degree of non-homogeneity of a motion vector of a block X and a motion vector of a neighboring block of the block X, which is obtained according to an equation

11. The image decoding method of claim 8, wherein the merged block is obtained by comparing the obtained variation and the predetermined threshold value, and merging the adjacent blocks included in the merged block when the obtained variation is lower than the predetermined threshold value.

14. A non-transitory computer-readable recording medium having recorded thereon a program for executing a method of claim 8.

12. An image decoding apparatus comprising:

an entropy decoder that is configured to obtain information about a merged block obtained by merging adjacent blocks from among blocks forming an image and about a motion vector of the merged block, from a received bitstream; and

a motion compensator that is configured to obtain an estimation value of the merged block by performing motion compensation on the merged block by using the motion vector,

wherein the merged block is obtained by merging the adjacent blocks based on a result of comparing a variation of a motion estimation error with respect to a homogeneity degree of a motion vector according to the merging of the adjacent blocks with a predetermined threshold value.

15. An image encoding method comprising:

obtaining a variation of a motion estimation error with respect to a homogeneity degree of a motion vector according to mergence with respect to a merged block, based on the motion vector of the merged block, motion vectors of adjacent blocks included in the merged block, motion vectors of neighboring blocks of the merged block, and motion vectors of neighboring blocks of the adjacent blocks included in the merged block;

determining a merged block having a smallest variation from among merged blocks on which the obtaining is performed; and

determining whether to merge blocks included in the determined merged block, based on the obtained variation.

16. The image encoding method of claim 15, wherein the obtaining of the variation comprises:

obtaining a homogeneity degree of a motion vector by using the motion vectors of the adjacent blocks included in the merged block, the motion vectors of the neighboring blocks of the merged block, and the motion vectors of the neighboring blocks of the adjacent blocks included in the merged block, wherein the homogeneity degree indicates a difference between a degree of non-homogeneity between the motion vector of the merged block and the motion vectors of the neighboring blocks of the merged block, and a degree of non-homogeneity between each of the motion vectors of the adjacent blocks included in the merged block and each of the motion vectors of the neighboring blocks of the adjacent blocks included in the merged block.

17. The image encoding method of claim 16, wherein the obtaining of the variation further comprises:

obtaining a variation of a motion estimation error, which indicates a difference between a motion estimation error by the motion vector of the merged block and a motion estimation error by each of the motion vectors of the adjacent blocks included in the merged blocks and

obtaining the variation of the motion estimation error with respect to the homogeneity.

18. The image encoding method of claim 15, further comprising encoding mergence information of the merged block in a bitstream.