Gradient-Based Position Dependent Prediction Combination for Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video coding technologies, such as HEVC and VVC, face limitations in predicting video blocks due to the lack of consideration for gradient information from neighboring samples, which affects prediction accuracy and coding efficiency.
Innovation Solution
The implementation of a gradient-based position-dependent prediction combination (PDPC) method that utilizes the gradient of neighboring samples to enhance prediction accuracy and flexibility during video encoding and decoding processes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional video coding technologies (HEVC, VVC) are used for predicting video blocks, then the coding process is simpler, but prediction accuracy deteriorates due to lack of gradient information from neighboring samples
Solution Approach 1:
The patent changes the parameters used in prediction from conventional reference sample values to gradient-based parameters. By computing gradients (differences) of neighboring samples and using these gradient values as prediction parameters, the method captures local variations more effectively, thereby improving prediction accuracy while maintaining manageable complexity through systematic gradient calculation and position-dependent combination rules.
Solution Approach 2:
The patent introduces gradient information as an intermediary element between neighboring samples and prediction values. Instead of directly using neighboring sample values for prediction, the method first computes gradients from these samples and then combines them position-dependently to form predictions. This intermediary gradient representation enables more accurate capture of local image characteristics.
2Manufacturing precision
If gradient-based position dependent prediction combination is applied to target block, then prediction quality improves, but computational complexity increases
Solution Approach 1:
The patent applies local quality by using position-dependent prediction combination where different weights are assigned to different positions within the target block. The gradient-based prediction values are combined with position-specific coefficients that vary across the block, allowing the prediction to adapt to local characteristics at each position rather than applying a uniform prediction approach throughout the entire block.
Solution Approach 2:
The patent implements a balanced approach by using a limited number of neighboring samples (e.g., a specific number of top and left samples) to compute gradients, rather than using all available samples. This partial action approach provides sufficient gradient information for accurate prediction while avoiding excessive computational burden that would result from processing all possible neighboring samples.
3Measurement precision
If more neighboring samples are used in gradient-based prediction combination, then prediction accuracy improves, but processing time increases
Solution Approach 1:
The patent uses a selective and limited subset of neighboring samples (a specific number of top and left samples) rather than all available neighboring samples. This partial action provides sufficient gradient information for accurate prediction while significantly reducing the number of computations required compared to using all neighboring samples, thus balancing accuracy with processing efficiency.
Data Source
AI summary
Embodiments of the present disclosure provide a method for video processing. The method comprises: applying, during a conversion between a target block of a video and a bitstream of the video, a gradient-based position dependent pre-diction combination to the target block in a coding mode, a gradient of a number of neighboring samples of the target block being used in the gradient-based position dependent prediction combination; and performing the conversion based on the applying.