Multiple Reference Line Decoding With Low-Memory Sub-Block Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding technologies face inefficiencies in intra prediction due to the increasing number of possible directions, leading to increased bit usage for less likely prediction directions, and inter prediction is limited by redundancy in motion vectors, which affects compression efficiency.
Innovation Solution
Implementing a method for multiple reference line intra prediction in video decoding, where a device extracts a parameter for a non-adjacent reference line and partitions the current block into sub-blocks, using a top adjacent reference line for boundary sub-blocks, and utilizing an apparatus with a processor to perform these methods.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If multiple reference lines are used for intra prediction, then prediction accuracy is improved, but memory requirements increase
Solution Approach 1:
The current block is divided into multiple sub-blocks, and different reference lines are selectively applied to different sub-blocks based on their spatial location. This segmentation allows the decoder to use multiple reference lines for improved prediction accuracy while limiting the memory buffer requirements by processing smaller sub-blocks independently rather than requiring all reference lines for the entire block.
Solution Approach 2:
Different reference lines are assigned to different sub-blocks based on their spatial characteristics and proximity to the reference lines. Sub-blocks closer to certain reference lines use those reference lines for prediction, optimizing local prediction quality while reducing overall memory requirements by not requiring all reference lines to be available for all sub-blocks.
2Adaptability or versatility
If more prediction directions are supported, then adaptability is improved, but bit usage increases
Solution Approach 1:
The prediction direction and reference line selection are dynamically adapted based on the specific characteristics of each sub-block and its position within the current block. Rather than supporting all possible directions uniformly across the entire block, the system dynamically selects appropriate reference lines for each sub-block, improving adaptability while reducing the bits needed to signal prediction parameters.
3Adaptability or versatility
If multiple reference lines are buffered for all blocks, then prediction flexibility is improved, but device complexity increases
Solution Approach 1:
The block processing is segmented into sub-blocks, each independently handled with appropriate reference line selection. This segmentation simplifies buffer management by limiting the scope of reference line requirements to smaller sub-blocks rather than requiring all reference lines to be buffered for potential use in any block.
Data Source
AI summary
An example method of video decoding includes receiving a coded video bitstream for a current block and partitioning the current block to obtain a plurality of sub-blocks. The method further includes, when a multiple reference line selection (MRLS) is applied for the current block, parsing a parameter from the coded video bitstream, the parameter indicating a reference line index corresponding to the MRLS. The method also includes, when a sub-block in the plurality of sub-blocks is located at a top boundary of the current block, using top-left values corresponding to an adjacent reference line to substitute values for top-left values of a non-adjacent reference line corresponding to the non-adjacent reference line index for the first sub-block. The method further includes, when the sub-block is not located at the top boundary, forgoing using the top-left values corresponding to the adjacent reference line to substitute values of a non-adjacent reference line.