Neighbor block-based disparity vector derivation in 3D-AVC

Patent metadata

Field	Value
Publication number	US-9237345-B2
Application number	US-201414189177-A
Country	US
Kind code	B2
Filing date	Feb 25, 2014
Priority date	Feb 26, 2013
Publication date	Jan 12, 2016
Grant date	Jan 12, 2016

Abstract

Techniques are described for deriving a disparity vector for a current block based on a disparity motion vector of a neighboring block in a 3D-AVC video coding process. The disparity vector derivation allows for texture-first coding where a depth view component of a dependent view is coded subsequent to the coding of the corresponding texture component of the dependent view.

First claim

What is claimed is:

1. A method of decoding video data for 3D-Advanced Video Coding (3D-AVC) comprising:
receiving a coded bitstream in a 3D-AVC compliant video coding process generated with texture-first coding of dependent views;
decoding a texture view component of a dependent view of the dependent views in the 3D-AVC compliant video coding process, wherein decoding the texture view component comprises:
evaluating motion information of one or more neighboring blocks of a current macroblock in the texture view component to determine whether at least one neighboring block of the current macroblock is inter-view predicted with a disparity motion vector that refers to an inter-view reference picture in a view other than the dependent view,
wherein evaluating the motion information of the one or more neighboring blocks of the current macroblock comprises:
prior to evaluating one or more spatial neighboring blocks of the current macroblock, evaluating one or more temporal neighboring blocks of the current macroblock in only up to two temporal pictures to determine whether at least one temporal neighboring block of the current macroblock is inter-view predicted with the disparity motion vector, wherein the at least one temporal neighboring block is in one of the two temporal pictures, wherein the two temporal pictures comprise a first temporal picture and a second temporal picture, wherein the first temporal picture comprises a picture identified at a first entry of a first reference picture list, and wherein the second temporal picture comprises a picture identified at a first entry of a second reference picture list;

deriving a disparity vector for the current macroblock based on the disparity motion vector for one of the evaluated neighboring blocks; and
assigning the derived disparity vector to each block within the current macroblock; and

decoding a depth view component that corresponds to the texture view component subsequent to decoding the texture view component.

2. The method of claim 1 , further comprising:
implementing one or more coding tools utilizing the derived disparity vector without using the depth view component for deriving the disparity vector.

3. The method of claim 1 , wherein evaluating motion information comprises:
evaluating motion information of only the one or more neighboring blocks, of the current macroblock, with motion vectors that refer to the first reference picture list of two reference picture lists to determine whether the at least one neighboring block of the current macroblock is inter-view predicted with the disparity motion vector; and
avoiding the evaluation of motion information of the one or more neighboring blocks, of the current macroblock, with motion vectors that refer to the second reference picture list of the two reference picture lists to determine whether at least one neighboring block of the current macroblock is inter-view predicted with the disparity motion vector.

4. The method of claim 1 , wherein evaluating one or more temporal neighboring blocks comprises:
evaluating motion information for only one bottom-right block in at least one of the first temporal picture or the second temporal picture, wherein the bottom-right block comprises a block located at a position bottom and to right of the current macroblock and in the at least one of the first temporal picture or the second temporal picture.

5. The method of claim 1 , wherein evaluating motion information further comprises:
evaluating motion information of up to a maximum of four spatial neighboring blocks after evaluating one or more temporal neighboring blocks.

6. The method of claim 5 , wherein three of the four spatial neighboring blocks comprise the same neighboring blocks used for motion prediction in the H.264/AVC video coding standard.

7. The method of claim 5 , wherein the four spatial neighboring blocks comprise:
a block located left of the current macroblock, a block located above the current macroblock, a block located above-left of the current macroblock, and a block located above-right of the current macroblock.

8. The method of claim 1 , wherein deriving a disparity vector comprises:
setting a horizontal component of the disparity vector equal to a horizontal component of the disparity motion vector; and
setting a vertical component of the disparity vector equal to zero.

9. A method of encoding video data for 3D-Advanced Video Coding (3D-AVC) comprising:
encoding a texture view component of a dependent view in a 3D-AVC compliant video coding process, wherein encoding the texture view component comprises:
evaluating motion information of one or more neighboring blocks of a current macroblock in the texture view component to determine whether at least one neighboring block of the current macroblock is inter-view predicted with a disparity motion vector that refers to an inter-view reference picture in a view other than the dependent view,
wherein evaluating the motion information of the one or more neighboring blocks of the current macroblock comprises:
prior to evaluating one or more spatial neighboring blocks of the current macroblock, evaluating one or more temporal neighboring blocks of the current macroblock in only up to two temporal pictures to determine whether at least one temporal neighboring block of the current macroblock is inter-view predicted with the disparity motion vector, wherein the at least one temporal neighboring block is in one of the two temporal pictures, wherein the two temporal pictures comprise a first temporal picture and a second temporal picture, wherein the first temporal picture comprises a picture identified at a first entry of a first reference picture list, and wherein the second temporal picture comprises a picture identified at a first entry of a second reference picture list;

deriving a disparity vector for the current macroblock based on the disparity motion vector for one of the evaluated neighboring blocks; and
assigning the derived disparity vector to each block within the current macroblock;

encoding a depth view component that corresponds to the texture view component subsequent to encoding the texture view component; and
generating for output a coded bitstream with texture-first coding of dependent views that includes the encoded texture view component and the encoded depth view component.

10. The method of claim 9 , further comprising:
implementing one or more coding tools utilizing the derived disparity vector without using the depth view component for deriving the disparity vector.

11. The method of claim 9 , wherein evaluating motion information comprises:
evaluating motion information of only the one or more neighboring blocks of the current macroblock with motion vectors that refer to the first reference picture list of two reference picture lists to determine whether at least one neighboring block of the current macroblock is inter-view predicted with the disparity motion vector; and
avoiding the evaluation of motion information of the one or more neighboring blocks of the current macroblock with motion vectors that refer to the second reference picture list of the two reference picture lists to determine whether at least one neighboring block of the current macroblock is inter-view predicted with the disparity motion vector.

12. The method of claim 9 , wherein evaluating one or more temporal neighboring blocks comprises:
evaluating motion information for only one bottom-right block in at least one of the first temporal picture or the second temporal picture, wherein the bottom-right block comprises a block located at a position bottom and to right of the current macrob

Assignees

• Qualcomm Inc

Inventors

• Kang Jewon
• Chen Ying
• Zhang Li

Classifications

• H04N19/597 Primary

  specially adapted for multi-view video sequence encoding · CPC title

• H04N19/513

  Processing of motion vectors · CPC title

• H04N19/56

  Motion estimation with initialisation of the vector search, e.g. estimating a good candidate to initiate a search · CPC title

• H04N19/577

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

• H04N19/52

  by predictive encoding · CPC title