Fused Multiple Reference Lines for Non-Square Video Block Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding standards like HEVC and VVC face challenges in efficiently predicting pixel blocks due to limitations in intra-prediction and cross-component prediction, particularly for non-square blocks with wide-angular directions and aspect ratios, leading to suboptimal compression efficiency.
Innovation Solution
The method employs multiple reference lines for predictive coding, including adjacent and non-adjacent lines, which are blended into a fused reference line for improved prediction, and utilizes techniques like decoder-side intra-mode derivation (DIMD), template-based intra-mode derivation (TIMD), cross-component linear model (CCLM), and multi-model CCLM (MMLM) to enhance prediction accuracy and efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional intra-prediction methods are used, then the prediction process is simple, but prediction accuracy is insufficient for non-square blocks with wide-angular directions
Solution Approach 1:
The patent segments the prediction process into multiple independent reference lines (first reference line, second reference line, etc.) that can be processed separately. Each reference line corresponds to a specific angular direction, allowing the complex wide-angular prediction to be broken down into simpler directional components that are easier to compute and combine.
Solution Approach 2:
The patent introduces an angular dimension to the prediction process by using multiple reference lines at different angles rather than a single reference line. This dimensional expansion from 1D (single reference line) to 2D/3D (multiple angular reference lines) enables accurate prediction for non-square blocks with wide-angular directions while maintaining computational simplicity through systematic angular sampling.
2Measurement precision
If multiple reference lines are used for prediction, then prediction accuracy improves, but computational complexity increases
Solution Approach 1:
The patent applies partial action by using a limited number of reference lines (first reference line, second reference line, etc.) rather than exhaustively using all possible reference lines. This selective approach provides sufficient prediction accuracy for practical purposes while significantly reducing computational complexity compared to using all possible reference lines.
Solution Approach 2:
The patent changes the parameter of reference line selection based on block characteristics (square vs. non-square, angular direction). By adapting the number and configuration of reference lines to the specific prediction needs, the system achieves high accuracy for difficult cases without unnecessarily increasing complexity for easier cases.
3Productivity
If conventional prediction methods are used, then processing is fast, but compression efficiency is suboptimal
Solution Approach 1:
The patent performs preliminary actions by pre-defining the reference line configurations and angular directions before the actual prediction process. This preparation allows the prediction to proceed quickly during encoding/decoding without requiring complex real-time calculations, as the reference framework is already established and can be directly applied to the current block.
Data Source
AI summary
A method using multiple reference lines for predictive coding is provided. A video coder receives data for a block of pixels to be encoded or decoded as a current block of a current picture of a video. The video coder receives or signals a selection of first and second reference lines among a plurality of reference lines that neighbor the current block. The video coder blends the first and second reference lines into a fused reference line. The video coder generates a prediction of the current block by using samples of the fused reference line. The video coder encodes or decodes the current block by using the generated prediction.