Video Decoder Resolution Scaling via Reference Picture Buffer
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video coding systems face inefficiencies when changing picture resolution, as they often require an intra-random access point (IRAP) picture, which increases data complexity, breaks inter-prediction chains, and strains decoder resources, especially during network congestion or buffer management issues.
Innovation Solution
The proposed solution involves dynamically changing picture resolutions without employing an IRAP picture by scaling decoded reference pictures within the decoded picture buffer, allowing inter-prediction to continue uninterrupted and reducing the need for additional data transmission and buffer refresh.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If an IRAP picture is employed to change resolution, then resolution change is achieved, but coding efficiency decreases and data size increases
Solution Approach 1:
The patent implements dynamic resolution scaling by allowing the decoded picture buffer to store reference pictures at different resolutions. The scaling process dynamically adjusts reference picture resolution based on current picture requirements, enabling flexible resolution changes without requiring IRAP pictures. This dynamic approach maintains inter-prediction continuity while adapting to varying resolution needs.
Solution Approach 2:
The patent changes the resolution parameter of decoded reference pictures through scaling operations. By modifying the resolution parameter dynamically without breaking inter-prediction chains, the system achieves resolution adaptation while maintaining coding efficiency. The scaling transforms reference pictures from one resolution to another, allowing seamless resolution transitions.
2Adaptability or versatility
If an IRAP picture is used for resolution change, then resolution adaptation is achieved, but network and processor resources are strained
Solution Approach 1:
The patent maintains continuous inter-prediction action across resolution changes by storing scaled reference pictures in the decoded picture buffer. This continuity eliminates the need for periodic IRAP picture insertions, reducing network transmission overhead and processor decoding efforts. The useful action of inter-prediction continues uninterrupted through resolution transitions.
Solution Approach 2:
The patent creates scaled copies of decoded reference pictures at different resolutions and stores them in the decoded picture buffer. These copied and scaled reference pictures enable resolution adaptation without requiring original high-resolution IRAP pictures, reducing network bandwidth consumption and processor resource usage while maintaining adaptation capability.
3Adaptability or versatility
If resolution changes are implemented frequently, then adaptability to network conditions improves, but decoder resource consumption increases
Solution Approach 1:
The patent performs preliminary scaling of decoded reference pictures and stores them in the decoded picture buffer before they are needed. This preliminary action prepares reference pictures at various resolutions in advance, allowing frequent resolution changes without requiring intensive real-time processing. The decoder resource consumption is reduced because scaling operations are performed proactively rather than reactively during playback.
Data Source
AI summary
A video coding mechanism is disclosed. The mechanism includes receiving a bitstream including a plurality of picture parameter sets (PPSs) and a plurality of pictures at a plurality of picture sizes. A first picture size of a first picture is determined from a first PPS. The first picture is decoded at the first picture size to create a decoded reference picture. The decoded reference picture is stored in a decoded picture buffer. A second picture size of a second picture is determined from a second PPS. The second picture references the decoded reference picture according to inter-prediction. The decoded reference picture is scaled from the first picture size to the second picture size. The second picture is decoded at the second picture size based on the decoded reference picture at the second picture size.