Subblock-Specific Intra-Inter Weights for Video Prediction Blocks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video compression technologies, such as VVC's combined inter-intra prediction (CIIP), face challenges in accurately predicting video blocks due to the use of a single weight for the entire block, leading to poorer performance in certain areas, particularly the bottom-right area, when using a weighted average of intra and inter prediction signals.
Innovation Solution
The method splits a current block into subblocks and applies different weight sets for each subblock, determining a weighted-sum prediction based on these subblock-specific weights for intra and inter prediction signals.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a single weight value is used for the entire prediction block in combined inter-intra prediction, then the prediction mode is simple and easy to implement, but the prediction accuracy deteriorates in certain areas particularly the bottom-right area
Solution Approach 1:
The prediction block is divided into multiple subblocks (e.g., top-left, top-right, bottom-left, bottom-right subblocks). Each subblock is then processed with its own weight values for intra and inter prediction signals, allowing different regions to have optimized prediction parameters rather than using a single weight for the entire block.
Solution Approach 2:
Different weight values are assigned to different subblocks based on their specific characteristics and prediction needs. For example, the bottom-right subblock may receive different weight combinations compared to the top-left subblock, allowing each region to have optimized local prediction quality tailored to its specific requirements.
2Measurement precision
If the PLANAR mode is always used for intra prediction regardless of neighboring block modes, then the prediction process is simplified, but the prediction performance deteriorates in the bottom-right area due to unavailable reference samples
Solution Approach 1:
The intra prediction mode is made dynamic rather than fixed. Instead of always using PLANAR mode, the system selects the appropriate intra prediction mode (PLANAR, angular, or DC mode) based on the prediction modes of neighboring blocks and the specific requirements of each subblock, allowing the prediction process to adapt to different conditions.
Solution Approach 2:
The prediction parameters including weight values and intra prediction modes are changed based on the position within the block and the modes of neighboring blocks. This allows the system to adjust parameters dynamically for each subblock to optimize prediction performance while managing computational complexity.
Data Source
AI summary
A method of generating a prediction block and a device or apparatus using same use a weighted-sum of an intra-prediction signal and an inter-prediction signal. The device or apparatus includes at least one processor configured to: determine that a current block of the video data is to be encoded in a weighted-sum prediction mode; split the current block into two or more subblocks; determine, for each of the subblocks, a weight set comprising a first weight for an intra-prediction signal of the current block and a second weight for an inter-prediction signal of the current block, the subblocks having different weight sets; and determine a prediction block of the current block by performing a weighted-sum of the intra-prediction signal and the inter-prediction signal on a per subblock basis, the weighted-sum being dependent on the weight set that is determined for each of the subblocks.