Method and apparatus of disparity vector derivation and inter-view motion vector prediction for 3D video coding

• US 9,736,498 B2
• Filed: 09/25/2013
• Issued: 08/15/2017
• Est. Priority Date: 10/03/2012
• Status: Active Grant

• Current Assignee: HFI Innovation, Inc. (MediaTek, Inc.)
• Original Assignee: MediaTek, Inc.
• Inventors: Chen, Yi-Wen, Lin, Jian-Liang
• Primary Examiner(s): Vaughn, Jr., William C
• Assistant Examiner(s): PEREZ FUENTES, LUIS M
• Application Number:
  US14/433,318
• Publication Number:
  US 20150264397A1
• Time in Patent Office: 1,420 Days
• Field of Search: 37524016
• CPC Class Codes: H04N 13/161 Encoding, multiplexing or demultiplexing different image signal components for multi-view video sequence encoding H04N19/597
• CPC Class Codes: H04N 19/176 the region being a block, e.g. a macroblock
• CPC Class Codes: H04N 19/52 by predictive encoding
• CPC Class Codes: H04N 19/573 Motion compensation with multiple frame prediction using two or more reference frames in a given prediction direction
• CPC Class Codes: H04N 19/597 specially adapted for multi-view video sequence encoding

Abstract

A method and apparatus for three-dimensional video coding and multi-view video coding are disclosed. Embodiments according to the present invention derive a unified disparity vector (DV) based on neighboring blocks of the current block or depth information associated with the current block and locate a single corresponding block in a reference view according to the unified DV. An inter-view motion vector prediction (MVP) candidate is then derived for both list0 and list1 from the single corresponding block. List0 and list1 MVs of the inter-view MVP candidate are derived from the single corresponding block located according to the unified DV.

Claims

1. A method for three-dimensional video coding and multi-view video coding, wherein reference pictures comprises reference pictures in a reference picture list 0 and reference pictures in a reference picture list 1, wherein the reference picture list 0 and the reference picture list 1 are associated with different prediction directions, the method comprising:

receiving input data associated with a current block in a dependent view, wherein the input data received at a video encoder side at least corresponds to one or more;

of original pixel data or depth data associated with the current block, and wherein the input data received at a video decoder side comprises a video bitstream corresponding to compressed data including the current block;

deriving a unified disparity vector (DV) based on at least one of;

disparity vectors (DVs) of neighboring blocks, according to a search order associated with the neighboring blocks;

disparity vector (DV) derived from depth information associated with the current block;

or

a linear combination or a statistic of selected DVs for the neighboring blocks, wherein the selected DVs are derived by checking available list0 DVs or available list1 DVs of the neighboring blocks;

locating a single corresponding block in a reference view according to the unified DV by pointing from a position of the current block to a position in the reference view through the unified DV to locate the single corresponding block;

deriving both list0 motion vector and list1 motion vector of an inter-view motion vector prediction (MVP) candidate from the single corresponding block in the reference view; and

encoding the input data using prediction information comprising the inter-view MVP candidate or decoding the input data using the prediction information comprising the inter-view MVP candidate.

2. The method of claim 1, wherein the neighboring blocks comprise a left block, an upper block and an upper-right block, and wherein the upper-right block is replaced by an upper-left block if the DV associated with the upper-right block is unavailable.

3. The method of claim 2, wherein the unified DV is derived from the DVs of the neighboring blocks according to a search order, and wherein the search order corresponds to searching the left block first, followed by the upper block, and then followed by the upper-right block.

4. The method of claim 1, wherein the unified DV is derived from the DVs of the neighboring blocks according to a search order.

5. The method of claim 1, wherein the unified DV is derived from the DVs of the neighboring blocks associated with list0 and list1.

6. The method of claim 5, wherein the DV of each neighboring block in one list is checked to find an available DV;

if no DV is available in said one list, the DV of said each neighboring block in another list is checked to find the available DV;

the available DV is designated as a selected DV of said each neighboring block; and

wherein said one list is selected from a group consisting of list0 and list1.

7. The method of claim 6, wherein if no available DV is found for said each neighboring block after checking the DV of said each neighboring block in both list0 and list1, a zero DV or a converted DV is used as the selected DV for said each neighboring block, wherein the converted DV is converted from one or more depth values, one or more neighboring texture or depth blocks of said each neighboring block.

8. The method of claim 5, wherein the unified DV is derived by first checking the DVs of the neighboring blocks in one list to find an available DV;

if no DV is available, the unified DV is derived by checking the DVs of the neighboring blocks in another list to find the available DV; and

wherein said one list is selected from a group consisting of list0 and list1.

9. The method of claim 8, wherein said checking the DVs of the neighboring blocks is terminated whenever one available DV is found.

10. The method of claim 8, wherein the DV of one neighboring block is counted as unavailable if the DV does not exist, or the corresponding block located by the DV of said one neighboring block is intra-coded or not available.

11. The method of claim 8, wherein if no available DV is found after checking the DVs of the neighboring blocks in both list0 and list1, a zero DV or a converted DV is used as the unified DV, wherein the converted DV is converted from one or more depth values associated with the current block.

12. The method of claim 1, wherein the unified DV is derived by converting one or more depth values associated with the current block.

13. The method of claim 12, wherein the unified DV is derived by converting the depth values from four corners associated with the current block.

14. An apparatus for three-dimensional video coding and multi-view video coding, wherein reference pictures comprises reference pictures in a reference picture list 0 and reference pictures in a reference picture list 1, wherein the reference picture list 0 and the reference picture list 1 are associated with different prediction directions, the apparatus comprising one or more electronic circuits and wherein said one or more electronic circuits are configured to:

receive input data associated with a current block in a dependent view, wherein the input data received at a video encoder side at least corresponds to one or more;

of original pixel data or depth data associated with the current block, and wherein the input data received at a video decoder side comprises a video bitstream corresponding to compressed data including the current block;

derive a unified disparity vector (DV) based on at least one of;

disparity vectors (DVs) of neighboring blocks, according to a search order associated with the neighboring blocks;

disparity vector (DV) derived from depth information associated with the current block;

or

a linear combination or a statistic of selected DVs for the neighboring blocks, wherein the selected DVs are derived by checking available list0 DVs or available list1 DVs of the neighboring blocks;

locate a single corresponding block in a reference view according to the unified DV by pointing from a position of the current block to a position in the reference view through the unified DV to locate the single corresponding block;

derive both list0 motion vector and list1 motion vector of an inter-view motion vector prediction (MVP) candidate from the single corresponding block in the reference view; and

encode the input data using prediction information comprising the inter-view MVP candidate or decode the input data using the prediction information comprising the inter-view MVP candidate.