Chroma Intra Prediction Using Non-Adjacent Reference Lines
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding technologies face challenges in achieving efficient compression of video data while maintaining video quality, particularly in adapting intra prediction modes for chroma components, leading to suboptimal coding efficiency.
Innovation Solution
Implement gradient-based intra mode derivation methods using multiple reference lines for decoder-side intra prediction, incorporating edge filters and constructing a unified predefined intra mode list to enhance prediction accuracy and diversity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If multiple reference lines with gradient-based intra mode derivation are used for chroma components, then coding efficiency is improved and prediction accuracy is enhanced, but device complexity and computational overhead increase
Solution Approach 1:
The patent applies segmentation by dividing the reference area into multiple reference lines (first reference line, second reference line, etc.) and performing gradient-based intra mode derivation separately for each reference line. This allows the system to process different reference lines independently, deriving different intra prediction modes for each, thereby improving prediction accuracy for chroma components while managing complexity through structured segmentation of the processing task.
2Measurement precision
If gradient-based intra mode derivation is performed on multiple reference lines, then intra prediction accuracy is improved, but computational overhead and processing time increase
Solution Approach 1:
The patent applies preliminary action by performing gradient-based intra mode derivation on multiple reference lines before final intra prediction. The system derives intra prediction modes from reference areas associated with each reference line in advance, preparing multiple candidate modes that can be selected based on prediction performance. This preliminary derivation of multiple modes allows for more accurate prediction while the selection process manages the computational overhead.
Data Source
AI summary
An example method of video decoding includes receiving a video bitstream comprising a plurality of blocks, including a current block. For each reference line of a plurality of reference lines for the current block: the method includes identifying a respective reference area corresponding to the reference line; and identifying a respective set of intra prediction modes by performing a gradient-based intra mode derivation on the respective reference area. The method includes populating a list of intra mode and reference line combinations using respective intra prediction modes from each respective set of intra prediction modes and a respective reference area according to a corresponding reference line; selecting an intra prediction mode and a reference line from the list of intra mode and reference line combinations; and decoding the current block using the intra prediction mode.