Reference Frame Context Averaging for Video Entropy Decoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding methods face challenges in achieving efficient entropy coding with accurate probability models, leading to suboptimal compression and increased signaling overhead.
Innovation Solution
Initializing the probability model for a current frame using a weighted average of contexts from multiple reference frames, employing fixed weights to enhance coding accuracy and reduce signaling overhead.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If default initialization values are used for probability models, then device complexity is reduced, but coding accuracy deteriorates
Solution Approach 1:
The patent applies preliminary action by pre-computing and storing probability models from reference frames before encoding the current frame. Instead of using default initialization values, the system prepares optimized probability models in advance by averaging contexts from multiple reference frames, which are then directly applied during current frame encoding to improve coding accuracy without increasing runtime complexity
Solution Approach 2:
The patent uses copying by replicating and adapting probability models from reference frames to the current frame. Specifically, contexts from multiple reference frames are copied and combined through weighted averaging to create the initial probability model for the current frame, eliminating the need for default initialization and improving accuracy through transferred knowledge
2Measurement precision
If multiple reference frames are used for probability model initialization, then coding accuracy is improved, but signaling overhead increases
Solution Approach 1:
The patent applies merging by combining probability models from multiple reference frames through weighted averaging. Instead of separately signaling probability models from each reference frame, the system merges them into a single composite probability model that captures the statistical characteristics of multiple references, thereby improving coding accuracy while avoiding redundant signaling overhead
Solution Approach 2:
The patent uses parameter changes by adjusting the weighting factors in the probability model averaging process. Different weights are assigned to different reference frames based on their relevance and quality, allowing the system to dynamically adapt the contribution of each reference frame to the current frame's probability model, optimizing both accuracy and signaling efficiency
Data Source
AI summary
An example method of video decoding includes receiving a video bitstream comprising a plurality of frames and identifying a plurality of reference frames for a current frame of the plurality of frames. The method also includes obtaining respective contexts for the plurality of reference frames and initializing a current frame context for the current frame by performing a weighted average of the respective contexts using a set of respective weights. The set of respective weights may comprise fixed weights. The method further includes entropy decoding one or more syntax elements for the current frame using the current frame context and reconstructing the current frame using the one or more syntax elements.