Bi-predictive merge mode based on uni-predictive neighbors in video coding

Patent metadata

Field	Value
Publication number	US-9648334-B2
Application number	US-201113336799-A
Country	US
Kind code	B2
Filing date	Dec 23, 2011
Priority date	Mar 21, 2011
Publication date	May 9, 2017
Grant date	May 9, 2017

Abstract

This disclosure describes a bi-predictive merge mode in which a bi-predictive video block inherits motion information from two different neighboring blocks, wherein the two different neighboring blocks were each encoded in a uni-predictive mode. Bi-predictive coding may improve the ability to achieve compression in video coding. The described bi-predictive merge mode may increase the number of bi-predictive candidates that can be used in the context of merge mode coding by allowing two separate uni-predicted neighbors to be used to define bi-predictive motion information for a video block.

First claim

The invention claimed is:

1. A method of using bi-predictive merge mode temporal prediction candidates for improving coding efficiency in decoding video data, the method comprising:
receiving one or more syntax elements for a current video block, wherein the one or more syntax elements define one or more merge mode candidates from a set of candidates to be used to code the current video block according to a bi-predictive merge mode that includes a bi-predictive merge mode candidate for improving the coding efficiency;
generating the set of candidates that includes the bi-predictive merge mode candidate for improving the coding efficiency, wherein a first candidate in the set of candidates is a left candidate that inherits motion information from a left neighboring block of a group of spatially neighboring video blocks of the current video block that includes multiple different spatially neighboring video blocks of the current video block that are positioned in a same frame of the encoded video data, a second candidate in the set of candidates is a temporal candidate that inherits motion information from a temporal neighbor block of the current video block that is positioned in a different frame with respect to the current video block, and a third candidate in the set of candidates is the bi-predictive merge mode candidate that inherits motion information from two different spatially neighboring video blocks of the group of spatially neighboring blocks for improving the coding efficiency, both of the two different spatially neighboring blocks being coded in uni-predictive mode;
determining that the one or more syntax elements indicate an index that defines the third candidate as being selected from the set of candidates for decoding the current video block; and
decoding the current video block based on the bi-predictive merge mode candidate that inherits the motion information from the two different spatially neighboring video blocks that are both coded in the uni-predictive modes.

2. The method of claim 1 , wherein decoding the current video block based on the bi-predictive merge mode candidate includes using the motion information inherited from the two different spatially neighboring video blocks of the current video block that are both coded in the uni-predictive modes, and wherein the inherited motion information comprises two different uni-predictive motion vectors associated with the two different spatially neighboring video blocks of the current video block.

3. The method of claim 2 , wherein the inherited motion information further comprises at least two values associated with the two different uni-predictive motion vectors, and wherein the at least two values identify one or more lists of predictive data associated with the two different uni-predictive motion vectors.

4. The method of claim 1 , wherein the current video block comprises a coding unit (CU) defined according to a high efficiency video coding (HEVC) standard, wherein the CU is defined relative to a largest coding unit (LCU) according to a quadtree partitioning scheme, the method further comprising:
receiving LCU syntax data that defines the quadtree partitioning scheme; and
receiving mode information for the CU that defines the bi-predictive merge mode candidate, wherein the one or more syntax elements are included in the mode information for the CU.

5. The method of claim 1 , wherein the current video block comprises a prediction unit (PU) of a coding unit (CU) that is defined according to a high efficiency video coding (HEVC) standard.

6. A method of temporal prediction for improving coding efficiency in encoding video data, the method comprising:
generating a bi-predictive merge mode candidate for encoding a current video block according to a bi-predictive merge mode, such that the bi-predictive merge mode candidate inherits motion information from two different spatially neighboring video blocks of the current video block, the two different spatially neighboring video blocks both being coded in uni-predictive modes, wherein the current video block and a group of spatially neighboring video blocks of the current video block that includes the two different spatially neighboring video blocks of the current video block are positioned in a same frame of the encoded video data;
selecting a candidate from a set of candidates for encoding the current video block according to the bi-predictive merge mode wherein a first candidate in the set of candidates is a left candidate that inherits motion information from a left neighboring block of the group of the spatially neighboring video blocks, a second candidate in the set of candidates is a temporal candidate that inherits motion information from a temporal neighbor block of the current video block that is positioned in a different frame with respect to the current video block, and a third candidate in the set of candidates is the bi-predictive merge mode candidate that inherits the motion information from the two different spatially neighboring video blocks that are coded in the uni-predictive modes, wherein the selection is based on rate-distortion analysis indicative of coding efficiency of the third candidate; and
upon selecting the third candidate from the set of candidates:
encoding the current video block based on the bi-predictive merge mode candidate that inherits the motion information from the two different spatially neighboring video blocks of the current block that are coded in the uni-predictive modes; and
generating one or more syntax elements that identify the bi-predictive merge mode candidate to a video decoder.

7. The method of claim 6 , wherein encoding the current video block based on the bi-predictive merge mode candidate includes using the motion information inherited from the two different spatially neighboring video blocks of the current video block that are coded in the uni-predictive modes and wherein the inherited motion information comprises two different uni-predictive motion vectors associated with the two different spatially neighboring video blocks of the current video block.

8. The method of claim 7 , wherein the inherited motion information further comprises at least two values associated with the two different uni-predictive motion vectors, wherein the at least two values identify one or more lists of predictive data associated with the two different uni-predictive motion vectors.

9. The method of claim 6 , wherein the current video block comprises a coding unit (CU) defined according to a high efficiency video coding (HEVC) standard, the method further comprising:
defining the CU relative to a largest coding unit (LCU) according to a quadtree partitioning scheme;
generating an LCU syntax data that defines the quadtree partitioning scheme; and
generating mode information for the CU that defines the bi-predictive merge mode candidate, wherein the one or more syntax elements are included in the mode information for the CU.

10. The method of claim 6 , wherein the current video block comprises a prediction unit (PU) of a coding unit (CU) that is defined according to a high efficiency video coding (HEVC) standard.

11. A video decoding device that uses bi-predictive merge mode temporal prediction candidates for improving coding efficiency to decode video data, the video decoding device comprising:
a memory configured to store at least a portion of the video data; and
a video decoder configured to:
receive one or more syntax elements for a current video block of the stored portion of the video data, wherein the one or more syntax elements define one or more merge mode candidates from a set of candidates to be used to code the current video block acco

Assignees

• Zheng Yunfei
• Chien Wei-Jung
• Karczewicz Marta
• Qualcomm Inc

Inventors

• Zheng Yunfei
• Chien Wei-Jung
• Karczewicz Marta

Classifications

• H04N19/52

  by predictive encoding · CPC title

• H04N19/105 Primary

  Selection of the reference unit for prediction within a chosen coding or prediction mode, e.g. adaptive choice of position and number of pixels used for prediction · CPC title

• H04N19/176

  the region being a block, e.g. a macroblock · CPC title

• H04N19/196 Primary

  being specially adapted for the computation of encoding parameters, e.g. by averaging previously computed encoding parameters (processing of motion vectors H04N19/513) · CPC title

• H04N19/577

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