ISP Video Coding Sub-Partition Width Constraints
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video coding techniques face challenges in achieving improved compression ratios with minimal sacrifice in picture quality, particularly in streaming and storage applications where bandwidth and memory resources are limited.
Innovation Solution
The method involves partitioning video blocks into sub-partitions for intra prediction, with each sub-partition having a width of at least 4 samples, and combining sub-partitions to maintain a combined width of at least 4 samples, allowing for efficient coding and decoding processes that optimize memory bandwidth and residual coefficients coding.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If video blocks are partitioned into fine sub-partitions for intra prediction, then prediction precision is improved, but memory bandwidth efficiency deteriorates and hardware compatibility becomes difficult
Solution Approach 1:
The video block is divided into sub-partitions for intra prediction, but the partitioning is constrained to produce sub-partitions with width >= 4 samples. This segmentation approach maintains prediction precision by allowing sub-partition division while preventing overly fine partitions that would harm memory efficiency.
Solution Approach 2:
The patent introduces a minimum width parameter (>= 4 samples) for sub-partitions to constrain the partitioning process. This parameter change ensures that while blocks can be divided for better prediction, the sub-partitions remain large enough to be memory-friendly and hardware-compatible.
2Measurement precision
If video blocks are partitioned into fine sub-partitions, then prediction accuracy is improved, but device complexity increases
Solution Approach 1:
The block is segmented into sub-partitions with controlled granularity (width >= 4), providing prediction accuracy improvements while maintaining hardware-friendly partition sizes that avoid excessive device complexity.
Solution Approach 2:
By setting a minimum width parameter of 4 samples for sub-partitions, the patent modifies the partitioning behavior to balance prediction accuracy with hardware compatibility, preventing the creation of overly complex partition structures.
3Measurement precision
If sub-partitions with small widths are used, then prediction detail is improved, but memory access efficiency deteriorates
Solution Approach 1:
The patent applies segmentation with a minimum width constraint (>= 4 samples) to create sub-partitions that provide sufficient prediction detail while maintaining memory access efficiency by avoiding overly narrow partitions.
Solution Approach 2:
The minimum width parameter of 4 samples is introduced to control sub-partition dimensions, ensuring that prediction detail is maintained through subdivision while memory access efficiency is preserved by preventing creation of narrow, memory-unfriendly partitions.
Data Source
AI summary
A method of video coding implemented by a coding device includes obtaining block size information of a current block to be coded in an intra sub-partition (ISP) coding mode, determining a partitioning direction based on the block size information, determining, based on the block size information, a number of sub-partitions to which the current block is to be partitioned in the partitioning direction, wherein the sub-partitions are for intra prediction separately with a width of each of the sub-partitions bigger than or equal to 4 or collaboratively by combining at least two of the sub-partitions with a combined width of the at least two sub-partitions bigger than or equal to 4, and coding the sub-partitions.