Gradient PDPC for Angular Intra-Prediction Video Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional video coding techniques struggle with applying position-dependent intra-prediction combination (PDPC) for angular intra-prediction modes when samples from the secondary boundary are not available, leading to inefficiencies in video compression.
Innovation Solution
The implementation of gradient PDPC, where a gradient term is calculated along the prediction direction to produce values for samples in the intra-prediction block, allowing for effective prediction even when secondary boundary samples are unavailable.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional PDPC is used for angular intra-prediction modes, then prediction accuracy is improved when secondary boundary samples are available, but PDPC must be disabled when secondary boundary samples are not available, leading to loss of compression efficiency
Solution Approach 1:
The patent introduces gradient terms as an intermediary mechanism to bridge the gap between primary boundary samples and the predicted block when secondary boundary samples are unavailable. Instead of directly using secondary boundary samples (which don't exist), the method calculates gradient terms from primary boundary samples to serve as a substitute, enabling PDPC to function in scenarios where it would traditionally be disabled.
Solution Approach 2:
The patent changes the parameter source for PDPC by transitioning from using actual secondary boundary samples to using calculated gradient terms. This parameter substitution allows the method to adapt to different boundary availability scenarios, maintaining PDPC functionality whether secondary samples are available or not by dynamically switching the source of correction values.
2Adaptability or versatility
If gradient PDPC is implemented to enable prediction when secondary boundary samples are unavailable, then adaptability of PDPC is improved, but computational complexity increases due to gradient term calculation
Solution Approach 1:
The patent applies gradient PDPC selectively rather than universally. Gradient term calculation is performed only when secondary boundary samples are unavailable, while conventional PDPC using actual secondary samples is used when they are available. This partial application approach enables adaptability without incurring the full computational cost of gradient calculations in all cases.
Solution Approach 2:
The patent performs preliminary checks to determine the availability of secondary boundary samples before deciding whether to calculate gradient terms. This preliminary action avoids unnecessary gradient calculations when secondary samples are available, thereby reducing computational complexity while maintaining the adaptability to handle cases where secondary samples are unavailable.
3Measurement precision
If gradient terms are calculated for all samples, then prediction accuracy is improved, but processing time and computational resources increase
Solution Approach 1:
The patent applies gradient PDPC with local quality by calculating gradient terms only for specific samples or regions where secondary boundary samples are unavailable, rather than uniformly for all samples. This localized application maintains prediction accuracy in critical areas while reducing overall processing time and computational resource consumption.
Data Source
AI summary
An example device for decoding video data includes a memory configured to store video data; and one or more processors implemented in circuitry and configured to: generate an intra-prediction block for a current block of video data using an angular intra-prediction mode, the angular intra-prediction mode being an upper-right angular intra-prediction mode or a lower-left angular intra-prediction mode; determine a prediction direction of the angular intra-prediction mode; for at least one sample of the intra-prediction block for the current block: calculate a gradient term for the at least one sample along the prediction direction; and combine a value of an intra-predicted sample of the intra-prediction block at a position of the at least one sample of the intra-prediction block with the gradient term to produce a value of the at least one sample of the intra-prediction block; and decode the current block using the intra-prediction block.