Sub-picture Extraction and Virtual Boundary Updates in Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding standards face decoding errors and quality issues when extracting sub-pictures with virtual boundaries outside the sub-picture, and do not adequately accommodate reference picture list constraints for pictures with multiple sub-pictures, leading to decoding mismatches.
Innovation Solution
Updating virtual boundaries for sub-pictures, modifying reference picture list constraints to handle sub-pictures predicted from inter-layer reference pictures, and ensuring sub-pictures are treated as pictures during reordering to improve decoding quality and performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If virtual boundaries are used for picture processing, then picture processing flexibility is improved, but decoding errors occur when virtual boundaries are outside the sub-picture
Solution Approach 1:
The patent makes the virtual boundary dynamic by allowing it to be adjusted based on sub-picture extraction requirements. When a sub-picture is extracted, the virtual boundary is dynamically repositioned to coincide with the sub-picture boundaries rather than remaining fixed at the original picture boundaries, thereby preventing decoding errors while maintaining processing flexibility.
Solution Approach 2:
The patent changes the parameter of virtual boundary position from a fixed value to an adjustable value. By modifying the virtual boundary position parameter to match sub-picture boundaries during extraction, the system resolves the conflict between maintaining flexibility for different processing scenarios and ensuring accurate decoding without boundary mismatches.
2Adaptability or versatility
If sub-pictures are extracted from pictures with virtual boundaries, then sub-picture extraction capability is improved, but decoding quality deteriorates due to boundary mismatches
Solution Approach 1:
The virtual boundary is made dynamic and adaptive to the extraction context. When sub-pictures are extracted, the virtual boundary automatically adjusts to align with the extracted sub-picture boundaries, ensuring that decoding operations always use appropriate boundaries and preventing quality degradation from boundary mismatches.
Solution Approach 2:
The patent applies different boundary characteristics to different regions. The virtual boundary position is locally adjusted to match each extracted sub-picture's boundaries rather than using a global fixed boundary, ensuring that each sub-picture is decoded with its own appropriate boundaries and maintaining high decoding quality.
3Productivity
If reference picture list constraints are applied to pictures with multiple sub-pictures, then reference picture management is improved, but decoding mismatches occur due to inadequate constraint accommodation
Solution Approach 1:
The patent segments the reference picture management by sub-picture. Reference picture list constraints are applied separately to each sub-picture rather than uniformly to the entire picture, allowing each sub-picture to have its own appropriate reference picture selections and preventing decoding mismatches while maintaining efficient management.
Solution Approach 2:
The patent modifies the reference picture list constraints parameters to accommodate multi-sub-picture structures. By adjusting the constraint parameters to recognize and handle sub-picture boundaries and relationships, the system achieves both efficient reference picture management and consistent decoding without mismatches.
Data Source
AI summary
Example techniques and devices for decoding video data are disclosed. An example device includes memory configured to store the video data and one or more processors implemented in circuitry and communicatively coupled to the memory. The one or more processors are configured to extract a current sub-picture from a bitstream of the video data and parse virtual boundary syntax elements indicative of virtual boundaries for a current picture, wherein the current sub-picture is associated with the current picture. The one or more processors are configured to update the virtual boundaries based at least in part on the virtual boundary syntax elements and decode the current sub-picture based on the updated virtual boundaries.