Adaptive Picture Size Signaling for Resolution-Aware Video Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding and decoding technologies face challenges in efficiently managing changes in picture size within a coded video sequence, particularly in scenarios where multiple semantically independent picture parts require adaptive resolution settings.
Innovation Solution
The method involves decoding candidate decoded resolutions from a parameter set, selecting a suitable resolution through an index in a transient header structure, resampling samples based on the output resolution and selected candidate decoded resolution, and enabling prediction using the resampled samples.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If picture size is changed within a coded video sequence to adapt to different activity levels, then compression efficiency is improved, but device complexity increases due to the need for resolution change signaling and processing
Solution Approach 1:
The patent applies parameter changes by dynamically adjusting the picture size (resolution) within a coded video sequence based on activity levels. The encoder determines activity levels for different picture parts and changes resolution parameters accordingly, allowing high-activity regions to use higher resolutions while low-activity regions use lower resolutions, thereby improving compression efficiency without uniformly increasing complexity across the entire sequence
Solution Approach 2:
The patent segments the video picture into multiple parts and applies different resolution settings to each segment based on its activity level. This segmentation allows independent resolution control for different regions, improving overall compression efficiency while managing complexity through localized processing rather than global resolution changes
2Adaptability or versatility
If multiple candidate resolutions are signaled in the parameter set to enable adaptive resolution change, then adaptability is improved, but loss of information increases due to the overhead of signaling resolution indices
Solution Approach 1:
The patent signals multiple candidate resolutions in the parameter set, but only actively uses and transmits data at the resolution determined by the activity level assessment. The additional candidate resolutions serve as pre-defined options that reduce signaling overhead compared to transmitting full resolution information dynamically, achieving a balance between adaptability and information loss
3Ease of operation
If reference pictures are resampled to different resolutions, then ease of operation is improved for handling diverse scene activities, but manufacturing precision deteriorates due to potential quality loss from resampling
Solution Approach 1:
The patent applies different resolution settings and resampling operations to different parts of the picture based on their activity levels. High-activity regions maintain higher resolution with minimal resampling, while low-activity regions use lower resolution with more aggressive resampling. This local quality approach ensures that picture quality is preserved where needed while allowing compression elsewhere
Data Source
AI summary
A method of decoding a coded picture of a coded video sequence is performed by at least one processor and the method includes decoding, from a parameter set, a plurality of candidate decoded resolutions, selecting, through an index coded in a transient header structure applicable to a group of samples, a candidate decoded resolution among the plurality of candidate decoded resolutions, resampling a sample of the group of samples based on an output resolution and the selected candidate decoded resolution, and enabling prediction using the resampled sample.