Wrap-Around Motion Compensation Disable for Video Encoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video coding technologies face increased implementation and computation complexity when using wrap-around motion compensation with reference picture resampling, especially when the reference picture width differs from the current picture width, leading to undesirable combinations of picture resolutions across layers and temporal pictures.
Innovation Solution
The method involves determining if the reference picture size equals the current picture size and disabling wrap-around motion compensation if they do not match, thereby encoding the current picture without using wrap-around motion compensation and generating an encoded video bitstream accordingly.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If wrap-around motion compensation is used with reference picture resampling when reference picture width differs from current picture width, then coding gain is improved, but implementation and computation complexity increases significantly
Solution Approach 1:
The patent changes the parameter of wrap-around offset determination from being relative to reference picture width to being relative to current picture width. This parameter change allows the wrap-around offset to remain consistent regardless of reference picture resolution, thereby reducing computation complexity while maintaining coding gain.
Solution Approach 2:
The patent segments the wrap-around motion compensation process into two distinct cases: one for when reference picture width equals current picture width (enabling wrap-around) and another for when they differ (disabling wrap-around). This segmentation simplifies the overall implementation by avoiding the need to handle complex width-matching scenarios.
2Object-affected harmful factors
If wrap-around motion compensation is used with reference picture resampling, then visual artifacts at seam lines are reduced, but adjustment of offset value according to picture width increases computation complexity
Solution Approach 1:
Instead of adjusting the wrap-around offset according to reference picture width (the conventional approach), the patent inverts the approach by determining the offset based on current picture width. This inversion eliminates the need for dynamic offset adjustment when reference pictures are resampled, thereby reducing computation complexity while still addressing visual artifacts.
3Adaptability or versatility
If reference picture resampling is performed to match different picture sizes, then inter-layer prediction with different picture sizes is enabled, but undesirable combinations of picture resolutions across layers and temporal pictures occur
Solution Approach 1:
The patent changes the fundamental parameter of how wrap-around offset is calculated from reference-picture-based to current-picture-based. This parameter change ensures that wrap-around motion compensation operates consistently regardless of reference picture resolution, thereby maintaining reliability across different picture resolutions while still enabling inter-layer prediction.
Data Source
AI summary
A method, device, and computer-readable medium for a method of encoding a video bitstream using at least one processor, including determining that reference picture resampling is enabled; determining a reference picture size of a reference picture; determining a current picture size of a current picture; determining whether the reference picture size equals the current picture size; based on determining that the reference picture size does not equal the current picture size, disabling wrap-around motion compensation; based on the wrap-around motion compensation being disabled, encoding the current picture without using the wrap-around motion compensation; and generating an encoded video bitstream based on the encoded current picture.