Video Intra Prediction With Non-Adjacent Reference Samples
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video signal encoding/decoding methods face inefficiencies in intra prediction, particularly when handling high-resolution and stereographic image content, leading to increased data transmission and storage costs.
Innovation Solution
The method involves partitioning a current block into multiple sub-blocks and performing intra prediction on each sub-block using specific reference sample lines, with shared intra prediction modes and non-adjacent reference samples derived from reconstructed samples.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If intra prediction is performed on the entire current block using conventional methods, then the processing is simple, but the prediction efficiency is insufficient for high-resolution video signals
Solution Approach 1:
The current block is divided into multiple sub-blocks, each of which is then processed independently through intra prediction. This segmentation allows the prediction process to handle smaller regions with more precision, improving overall prediction efficiency for high-resolution video signals while maintaining manageable computational complexity through systematic division.
2Measurement precision
If reference samples are selected from adjacent lines only, then the reference sample derivation is simple, but the prediction accuracy is limited
Solution Approach 1:
The reference sample derivation process extends from one-dimensional adjacent line selection to two-dimensional sampling by incorporating both horizontal and vertical reference sample lines. This dimensional expansion allows the system to access reference samples from multiple directions, significantly improving prediction accuracy while the structured approach to sample selection keeps the derivation process manageable.
3Productivity
If the same reference sample line is used for all sub-blocks, then the processing is simplified, but the prediction optimization is insufficient
Solution Approach 1:
Different sub-blocks are assigned different reference sample lines based on their specific positions and characteristics within the current block. This local customization allows each sub-block to utilize the most appropriate reference samples for its particular location, optimizing prediction efficiency while the automated assignment process maintains operational simplicity through systematic rules.
Data Source
AI summary
According to the present invention, there is provided a method of decoding an image, the method including: partitioning a current block into multiple sub-blocks; performing intra prediction on a first sub-block among the multiple sub-blocks by using a first sub reference sample line; and performing intra prediction on a second sub-block among the multiple sub-blocks by using a second sub reference sample line. Herein, an intra prediction mode of the first sub-block and an intra prediction mode of the second sub-block are the same.