Video Decoding with CCP Merge for Chroma Block Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding technologies inadequately predict chroma blocks using prediction models derived from neighboring blocks, leading to inefficient chroma block reconstruction.
Innovation Solution
A method and device for predicting a block unit using a cross-component prediction (CCP) merge list, combining intra prediction modes with weighted prediction blocks, and employing various CCP merge candidates including spatial, temporal, and history-based models to enhance chroma block prediction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If prediction models are derived from other reconstructed blocks, then the prediction process can be simplified, but the prediction precision deteriorates because the models are not derived from multiple neighboring samples
Solution Approach 1:
The patent segments the prediction process into two distinct stages: (1) deriving prediction models from multiple neighboring blocks using multiple neighboring samples, and (2) refining these models using model refinement modes. This segmentation allows the system to maintain comprehensive model derivation while adding optional refinement steps, thereby improving prediction precision without excessively complicating the basic prediction process.
Solution Approach 2:
The patent performs preliminary derivation of prediction models from multiple neighboring blocks before the actual prediction step. By pre-computing and storing these models in a buffer, the system prepares high-quality prediction data in advance, which then can be used directly during decoding without real-time complex calculations, thus maintaining both precision and efficiency.
2Measurement precision
If multiple model refinement modes are implemented, then prediction precision improves, but device complexity increases
Solution Approach 1:
The patent implements dynamic model refinement where the refinement process is adaptively applied based on specific conditions and requirements. Different model refinement modes can be selectively activated depending on the prediction scenario, allowing the system to adjust its complexity level dynamically rather than always operating at maximum refinement, thus balancing precision improvement with complexity management.
Solution Approach 2:
The patent changes key parameters of the prediction model during the refinement process, such as adjusting model coefficients and weights based on neighboring sample statistics. By dynamically modifying these parameters rather than using fixed model structures, the system achieves higher prediction precision while keeping the refinement mechanism relatively simple and computationally efficient.
3Speed
If prediction models are derived from single neighboring blocks, then the processing speed improves, but the prediction accuracy deteriorates due to inadequate coverage of chroma samples
Solution Approach 1:
The patent merges prediction models from multiple neighboring blocks into a combined prediction model for the current block. By combining information from multiple sources (top, left, and other neighboring blocks) rather than relying on a single block, the system achieves more accurate chroma prediction while maintaining efficient processing through the merging operation, thus resolving the contradiction between speed and accuracy.
Data Source
AI summary
A method of decoding video data performed by an electronic device is provided. The method receives the video data and determines a block unit from a current frame included in the video data. The method further determines an intra prediction mode from a plurality of intra default modes, determine a cross-component prediction (CCP) merge list of the block unit including a plurality of CCP merge candidates; and selecting one of the CCP merge candidates for the block unit to determine a prediction model of the selected CCP merge candidate. The method then predicts the block unit using the prediction model of the selected CCP merge candidate to generate a first prediction block, predicts the block unit based on the intra prediction mode to generate a second prediction block, and reconstructs the block unit based on the first prediction block and the second prediction block.