Wrap-around padding for 360-degree video seam artifacts
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
360-degree video mapped onto 2D surfaces using 3D-to-2D projection methods results in seam visual artifacts during re-projection and stitching of independently coded regions, which existing video coding and decoding techniques fail to adequately address.
Innovation Solution
The method involves decoding picture partitioning information to determine if padding is applied, and if so, identifying whether it includes wrap-around padding, applying either repetition padding or wrap-around padding accordingly to the sub-regions, and reconstructing the current picture based on the decoded sub-regions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If 360-degree video is mapped onto 2D surfaces using 3D-to-2D projection methods, then the video can be encoded and decoded by traditional 2D video coders, but seam visual artifacts are resulted from re-projection and stitching of independently coded regions
Solution Approach 1:
The patent applies preliminary action by introducing wrap-around padding during the encoding stage to pre-compensate for the seam artifacts that will occur during re-projection. The padding is applied to the left and right sides of the equirectangular projection, with the right side wrapped around to the left side, so that when the padded regions are stitched together during decoding, the seams are minimized. This advance preparation resolves the contradiction by enabling traditional 2D coding while mitigating the harmful seam artifacts.
Solution Approach 2:
The wrap-around padding acts as an intermediary element between the independently coded left and right regions. By introducing this padding layer that is wrapped around, the patent creates a transition zone that blends the boundaries between regions, reducing the visual discontinuities. The padding serves as a mediator that allows traditional 2D coding to be used while eliminating the harmful seam effects.
2Object-affected harmful factors
If padding is applied to sub-regions of the current picture, then seam artifacts are reduced, but the complexity of determining and applying the correct padding type increases
Solution Approach 1:
The patent implements feedback by using picture partitioning information that is decoded from the bitstream to determine the padding configuration. The decoder reads padding information from the parameter sets or slice headers, which specifies whether wrap-around padding should be applied and to which sub-regions. This feedback mechanism allows the decoder to adaptively apply the correct padding type based on the encoded picture partitioning, reducing seam artifacts while managing complexity through structured information flow.
Solution Approach 2:
The patent applies segmentation by dividing the picture into multiple sub-regions (such as slices or tiles) and applying different padding strategies to different segments based on their position and content. The picture partitioning information indicates which sub-regions require wrap-around padding and which require standard padding, allowing the decoder to handle each segment appropriately without uniformly increasing complexity across the entire picture.
Data Source
AI summary
A method of encoding a current picture includes based on a first determination indicating that the padding is not applied, encoding the plurality of sub-regions without padding; based on the first determination indicating that the padding is applied, making a second determination that the padding comprises wrap-around padding; based on the second determination indicating that the padding does not comprise wrap-around padding, applying repetition padding to the plurality of sub-regions; based on the second determination indicating that the padding comprises the wrap-around padding, obtaining an offset value from the picture partitioning information and encoding the plurality of sub-regions based on the wrap-around padding and the offset value; generating picture partitioning information based on the first determination and the second determination; and encoding the current picture based on the encoded plurality of sub-regions and the picture partitioning information.