Inter prediction method and apparatus in image coding system

• US 10,575,011 B2
• Filed: 08/25/2016
• Issued: 02/25/2020
• Est. Priority Date: 09/24/2015
• Status: Active Grant

• Current Assignee: LG Electronics, Inc. (LG Corporation)
• Original Assignee: LG Electronics, Inc. (LG Corporation)
• Inventors: Park, Naeri, Lim, Jaehyun, Seo, Jungdong, Nam, Junghak
• Primary Examiner(s): Anyikire, Chikaodili E
• Application Number:
  US15/761,665
• Publication Number:
  US 20180352247A1
• Time in Patent Office: 1,279 Days
• Field of Search: None
• CPC Class Codes: H04N 19/513 Processing of motion vectors
• 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 19/577 Motion compensation with bidirectional frame interpolation, i.e. using B-pictures

Abstract

An inter prediction method performed by a decoding apparatus according to the present invention comprises the steps of: receiving information associated with an MVD through a bit stream; deriving a candidate motion information list on the basis of a neighboring block of a current block; deriving an MVP of the current block on the basis of the candidate motion information list; deriving a motion vector of the current block on the basis of the MVP and the MVD; and generating a prediction sample with respect to the current block on the basis of the motion vector. According to the present invention, a motion vector may be derived on the basis of a candidate motion information list derived on the basis of a neighboring block. Therefore, the amount of data of prediction mode information can be reduced and inter prediction accuracy and overall coding efficiency can be improved.

Claims

1. An inter prediction method performed by a decoding apparatus, the method comprising:

receiving information on a motion vector difference (MVD) through a bit-stream;

constructing a motion vector predictor (MVP) candidate list based on neighboring blocks of a current block;

deriving a MVP of the current block based on the MVP candidate list;

deriving a motion vector of the current block based on the MVP and the MVD; and

generating a predicted sample with respect to the current block based on the motion vector,

wherein constructing the MVP candidate list comprises;

deriving temporary MVP candidates based on the neighboring blocks;

for a temporary MVP candidate which is an uni-predicted candidate, deriving a bi-predicted candidate by using the uni-predicted candidate, wherein the uni-predicted candidate includes one of a L0 motion vector or a L1 motion vector, and the bi-predicted candidate includes both of the L0 motion vector and the L1 motion vector; and

constructing the MVP candidate list of the current block including the bi-predicted candidate,

wherein Sum of Absolute Differences (SAD) between a L0 reference block for a MVP L0 motion vector for the MVP and a L1 reference block for a MVP L1 motion vector for the MVP is smaller than SAD between a L0 reference block for a non-MVP L0 motion vector for one of the temporary MVP candidates except the MVP and a L1 reference block for a non-MVP L1 motion vector for one of the temporary MVP candidates except the MVP,

wherein the MVP candidate list is constructed by replacing the uni-predicted candidate among the temporary MVP candidates with the bi-predicted candidate, and

wherein the number of the temporary MVP candidates is same with the number of all MVP candidates of the constructed MVP candidates list after the replacing.

2. The method of claim 1, wherein the L1 motion vector has a direction opposite to a direction of the L0 motion vector.

3. The method of claim 1, wherein, when the uni-predicted candidate includes the L0 motion vector, the L1 motion vector is derived by scaling the L0 motion vector.

4. The method of claim 3,

wherein the L1 motion vector is derived by scaling the L0 motion vector based on a first difference and a second difference,

wherein the first difference represents a difference between a picture order count (POC) of a current picture and a POC of a L0 reference picture associated with the L0 motion vector among reference pictures included in a reference picture list L0, and

wherein the second difference represents a difference between a POC of a L1 reference picture included in a reference picture list L1 and the POC of the current picture.

5. The method of claim 1,

wherein when the uni-predicted candidate includes the L0 motion vector, a L1 reference picture index associated with the L1 motion vector indicates a L1 reference picture of which POC difference with respect to a current picture is a specific POC difference among reference pictures included in a reference picture list L1, and

wherein the specific POC difference is identical to a POC difference between the POC of the current picture and a POC of a L0 reference picture associated with the L0 motion vector among reference pictures included in the reference picture list L0.

6. The method of claim 1, wherein when the uni-predicted candidate includes the L0 motion vector, a value of a L1 reference picture index associated with the L1 motion vector is derived to be a value identical to a minimum value among values of L1 reference picture indexes for the temporary MVP candidates.

7. The method of claim 1, wherein, when the uni-predicted candidate includes the L0 motion vector, a value of a L1 reference picture index associated with the L1 motion vector is derived to be a most probable value among values of L1 reference picture indexes for the temporary MVP candidates.

8. The method of claim 1,

wherein a value of a reference picture index associated with the L1 motion vector for the bi-predicted candidate indicates a second reference picture,

wherein the L1 motion vector for the bi-predicted candidate is derived by scaling the L0 motion vector for the bi-predicted candidate to an opposite direction with a direction of the L0 motion vector for the bi-predicted candidate based on the second reference picture,

wherein SAD between a first reference area indicated by the L0 motion vector for the bi-predicted candidate on a first reference picture and a second reference area indicated by the L1 motion vector for the bi-predicted candidate on the second reference picture is smaller than SAD between the first reference area and a third reference area on a third reference picture indicated by a L1-direction motion vector,

wherein the third reference picture is not equal to the second reference picture, and

wherein the L1-direction motion vector is derived by scaling the L0 motion vector for the bi-predicted candidate to the opposite direction with the direction of the L0 motion vector for the bi-predicted candidate.

9. The method of claim 1, wherein, when the uni-predicted candidate includes the L0 motion vector, a L1 motion vector is derived based on L1 motion vectors included in the temporary MVP candidates.

10. The method of claim 1, wherein, when the uni-predicted candidate includes the L0 motion vector, a L1 motion vector is derived by scaling a L0 motion vector, which is derived based on L0 motion vectors except for the L0 motion vector in the uni-predicted candidate from the temporary MVP candidates, in a L1 direction.

11. An inter prediction method performed by an encoding device, the method comprising:

deriving a motion vector predictor (MVP) candidate list based on neighboring blocks of a current block;

determining a MVP of the current block based on the MVP candidate list;

deriving a motion vector difference (MVD) of the current block based on the MVP; and

encoding video information including information on the MVD,

wherein constructing the MVP candidate list comprising;

deriving temporary MVP candidates based on the neighboring blocks;

for a temporary MVP candidate which is an uni-predicted candidate, deriving a bi-predicted candidate by using the uni-predicted candidate, wherein the uni-predicted candidate includes one of a L0 motion vector or a L1 motion vector, and the bi-predicted candidate includes both of the L0 motion vector and the L1 motion vector; and

constructing the MVP candidate list of the current block including the bi-predicted candidate,

wherein Sum of Absolute Differences (SAD) between a L0 reference block for a MVP L0 motion vector for the MVP and a L1 reference block for a MVP L1 motion vector for the MVP is smaller than SAD between a L0 reference block for a non-MVP L0 motion vector for one of the temporary MVP candidates except the MVP and a L1 reference block for a non-MVP L1 motion vector for one of the temporary MVP candidates except the MVP,

wherein the MVP candidate list is constructed by replacing the uni-predicted candidate among the temporary MVP candidates with the bi-predicted candidate, and

wherein the number of the temporary MVP candidates is same with the number of all MVP candidates of the constructed MVP candidates list after the replacing.