Motion compensation method and device using bidirectional optical flow
地域：
Seoul
摘要
Disclosed herein is a method for adaptive bidirectional optical flow estimation for inter-screen prediction compensation during video encoding. The method aims to reduce complexity and/or cost of bidirectional optical flow (BIO) at a pixel level or a subblock level.
說明書
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
This present application is a continuation of U.S. patent application Ser. No. 16/642,164, filed on Feb. 26, 2020, which is a national stage filing under 35 U.S.C. § 371 of PCT application number PCT/KR2018/009940 filed on Aug. 29, 2018 which is based upon and claims the benefit of priorities to Korean Patent Application No. 10-2017-0109632, filed on Aug. 29, 2017 and Korean Patent Application No. 10-2017-0175587, filed on Dec. 19, 2017, in the Korean Intellectual Property Office, which are incorporated herein in their entireties by reference.
The present disclosure relates to image encoding or decoding. More specifically, the present disclosure relates to a bidirectional optical flow for motion compensation.
The statements in this section merely provide background information related to the present disclosure and may not constitute prior art.
權(quán)利要求
What is claimed is:1. A video encoding method for predicting a target block using a bi-directional optical flow, the method comprising:determining the target block, by partitioning a pre-determined size of block in a tree structure; generating a first motion vector for a first reference picture and a second motion vector for a second reference picture; deriving a variable from sample value differences between samples in the first reference picture which are determined based on the first motion vector and samples in the second reference picture which are determined based on the second motion vector, wherein the variable is SAD (Sum of Absolute Differences) or SSE (Sum of Squared Errors) and is used for determining whether to apply the bi-directional optical flow; generating a prediction block of the target block from the first and second reference pictures, by selectively either applying or skipping the bi-directional optical flow depending on the variable; and encoding partition information related to the tree structure, and residual signals that are differences between samples in the target block and samples in the prediction block, wherein the bi-directional optical flow is performed by the unit of samples in the target block or by the unit of sub-blocks partitioned from the target block. 2. The apparatus of claim 1 , wherein the bi-directional optical flow is skipped when the variable is less than a predetermined threshold, and is applied when the variable is greater than the predetermined threshold. 3. The apparatus of claim 1 , wherein, when at least one of a width or a height of the target block is less than a predetermined length, the bi-directional optical flow is skipped. 4. The apparatus of claim 3 , wherein the predetermined length is 8. 5. A video decoding apparatus for predicting a target block using a bi-directional optical flow, the apparatus comprising one or more processor configured to:decode partition information from a bitstream, and determine the target block by partitioning a pre-determined size of block in a tree structure; reconstruct residual signals corresponding to the target block from the bitstream; generate a first motion vector for a first reference picture and a second motion vector for a second reference picture; derive a variable from sample value differences between samples in the first reference picture which are determined based on the first motion vector and samples in the second reference picture which are determined based on the second motion vector, wherein the variable is SAD (Sum of Absolute Differences) or SSE (Sum of Squared Errors) and is used for determining whether to apply the bi-directional optical flow; generate a prediction block of the target block from the first and second reference pictures, by selectively either applying or skipping the bi-directional optical flow depending on the variable; and reconstruct the target block by using samples in the prediction block and the residual signals, wherein the bi-directional optical flow is performed by the unit of samples in the target block or by the unit of sub-blocks partitioned from the target block. 6. The method of claim 5 , wherein the bi-directional optical flow is skipped when the variable is less than a predetermined threshold, and is applied when the variable is greater than the predetermined threshold. 7. The method of claim 5 , wherein, when at least one of a width or a height of the target block is less than a predetermined length, the bi-directional optical flow is skipped. 8. The method of claim 7 , wherein the predetermined length is 8. 9. A non-transitory recording medium storing a bitstream generated by a video encoding method for predicting a target block using a bi-directional optical flow, the method comprising:determining the target block, by partitioning a pre-determined size of block in a tree structure; generating a first motion vector for a first reference picture and a second motion vector for a second reference picture; deriving a variable from sample value differences between samples in the first reference picture which are determined based on the first motion vector and samples in the second reference picture which are determined based on the second motion vector, wherein the variable is SAD (Sum of Absolute Differences) or SSE (Sum of Squared Errors) and is used for determining whether to apply the bi-directional optical flow; generating a prediction block of the target block from the first and second reference pictures, by selectively either applying or skipping the bi-directional optical flow depending on the variable; and encoding partition information related to the tree structure, and residual signals that are differences between samples in the target block and samples in the prediction block, wherein the bi-directional optical flow is performed by the unit of samples in the target block or by the unit of sub-blocks partitioned from the target block.
意見反饋
使用該功能遇到問題