Video Coding Context Model Prediction Using Reference CTU Split Information
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video coding technologies face challenges in efficiently modeling context for block structure signaling, particularly in determining split flags for video encoding and decoding, which affects coding efficiency and compression ratios.
Innovation Solution
The proposed solution involves determining a context model for split flags based on split information from a reference coding tree unit (CTU), using spatial or temporal neighboring CTUs to predict the block structure, and employing context modeling techniques to select appropriate context models for split flags, such as quad tree and binary/extended quad tree splits, to improve coding efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional context modeling methods are used for split flag signaling, then the implementation is simpler, but coding efficiency is lower and compression ratios are worse
Solution Approach 1:
The patent applies preliminary action by determining the context model for split flags before actual decoding occurs. The method pre-establishes context models based on reference CTU split information and current CTU characteristics, so that when split flags are decoded, the appropriate context model is already ready, improving coding efficiency without adding complex runtime processing
Solution Approach 2:
The patent segments the context modeling process into distinct cases: one path for when the current CTU uses the same block structure as the reference CTU (reusing reference context models), and another path for when they differ (determining new context models). This segmentation allows the system to simplify modeling when possible while maintaining accuracy when needed
2Loss of information
If more split flags are explicitly signaled, then block structure information is more accurate, but bitstream size increases reducing compression ratio
Solution Approach 1:
The patent uses feedback by comparing the actual block structure of the current CTU with the reference CTU structure. Based on this comparison feedback, the system determines whether to reuse reference context models or establish new ones, allowing efficient signaling that adapts to actual content characteristics rather than always using explicit signaling
Solution Approach 2:
The patent changes the parameter being signaled from explicit split flag values to context model indices. Instead of directly signaling detailed block structure information, the system signals which pre-determined context model to use, reducing the number of bits required while maintaining the ability to represent various block structures through the context model selection
Data Source
AI summary
Aspects of the disclosure provide methods and apparatuses for video encoding/decoding. In some examples, an apparatus for video decoding includes processing circuitry. The processing circuitry receives coded information of a current coding tree unit (CTU) from a coded video bitstream. Then, the processing circuitry determines a context model for a split flag associated with a current block within the current CTU at least partially based on split information of a corresponding block in a reference CTU for the current CTU. The split flag associated with the current block is indicative of split information of the current block. Then, the processing circuitry determines the split flag based on the context model, and decodes the current block based on the split flag that is determined based on the context model.