Weighted Angular Prediction Using Multi-Reference Lines
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video coding standards, such as HEVC, face limitations in coding efficiency for higher resolutions and bit-rates, particularly in angular prediction methods that rely on single reference lines, leading to potential noise propagation and reconstruction quality issues.
Innovation Solution
The method involves generating predictor pixels by combining projected pixel values from both a top reference row and a left reference column, with weights determined from a weighting table based on distance, to improve the accuracy of angular prediction in JVET video coding.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If single reference line angular prediction is used, then device complexity is reduced, but manufacturing precision deteriorates due to noise propagation and reconstruction quality issues
Solution Approach 1:
The prediction process is segmented into multiple independent reference lines (top reference row and left reference column) instead of using a single reference line. Each reference line provides independent prediction values that are subsequently combined, allowing the system to distribute the prediction task across multiple segments to improve accuracy while managing complexity
Solution Approach 2:
Prediction values from multiple reference lines are merged through a combination process where projected pixel values from both the top reference row and left reference column are integrated. This merging of multiple prediction sources improves reconstruction quality by reducing noise propagation inherent in single-reference-line methods
2Manufacturing precision
If multiple reference lines are combined with weighting, then manufacturing precision is improved through better prediction accuracy, but device complexity increases due to additional computation and memory requirements
Solution Approach 1:
Different regions of the prediction block utilize different reference lines and weighting schemes based on their local characteristics. The weighting factors are adjusted locally according to the distance from reference pixels, allowing the system to optimize prediction accuracy for each local region while managing overall complexity through localized rather than global processing
Solution Approach 2:
Weighting factors are dynamically changed based on distance parameters between predictor pixels and projected pixel positions. By adjusting the weighting parameters according to spatial distance, the system improves prediction accuracy without requiring complex adaptive algorithms, as the parameter changes follow a predictable distance-based pattern
Data Source
AI summary
A method and apparatus for decoding JVET video, including receiving a bitstream and generating predictor pixels for angular prediction using pixels at projected positions along both a top reference row (main reference line) and a left reference column (side reference line). By combining projected pixel values on the main reference line with projected pixel values on the side reference, a predictor pixel at coordinate (x,y) can be determined. Further weighting the values according to a distance between predictor pixels and projected pixel positions on the main and side references may be included in the combination of pixel values. The weight parameter may be determined from a weighting table. Further, the weights for horizontal and vertical predictors may be computed based on that of a horizontal and vertical predictor, whichever is more accurate.