Intra Sub-partitioning Video Coding Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current video coding technologies face challenges in efficiently managing intra sub-partitioning, wide-angle prediction, delta quantization, and transform operations, leading to suboptimal compression performance and increased bandwidth usage.
Innovation Solution
The implementation of intra sub-partitioning (ISP) mode in video coding, which allows for partitioning of video blocks into sub-partitions, enables wide-angle prediction, and adjusts delta quantization parameters, while also determining the applicability of transform operations based on block characteristics and coding tools.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If intra sub-partitioning mode is used to partition video blocks into multiple sub-partitions, then prediction accuracy is improved, but device complexity and processing overhead increase
Solution Approach 1:
The video block is divided into multiple sub-partitions (e.g., 2x2 sub-partitions) with different intra prediction modes applied to each sub-partition independently. This segmentation allows each sub-region to be predicted with the most appropriate mode, improving overall prediction accuracy while managing complexity through structured subdivision.
Solution Approach 2:
Different intra prediction modes are assigned to different sub-partitions based on local characteristics. For example, planar mode may be applied to one sub-partition while angular modes are applied to others, allowing each region to have optimized prediction quality tailored to its specific content characteristics.
2Measurement precision
If wide-angle intra prediction modes are added to handle obtuse angle cases, then prediction performance for wide-angle blocks is improved, but device complexity increases
Solution Approach 1:
The patent introduces wide-angle prediction modes that extend the traditional intra prediction mode range to handle obtuse angle cases. By adding new prediction modes with specific angle parameters (e.g., modes 65-90 for wide-angle predictions), the system can accurately predict blocks with wide-angle geometries while maintaining a systematic approach to mode selection and management.
3Productivity
If transform operations are applied to all coding units, then compression performance is improved, but processing time and energy consumption increase
Solution Approach 1:
Transform operations are selectively applied only to certain coding units or sub-partitions based on specific criteria (e.g., residual energy, block size, prediction mode). By applying transforms partially rather than universally, the system achieves good compression performance for regions that benefit most while reducing overall processing time and energy consumption.
Solution Approach 2:
The patent extracts and applies transform operations only where necessary by evaluating specific conditions for each coding unit. Coding units that meet certain criteria (such as having significant residual energy or specific prediction modes) undergo transformation, while others are excluded from the transform process, optimizing the balance between compression and processing cost.
4Measurement precision
If delta quantization parameters are adjusted for different sub-partitions, then quantization precision is improved, but device complexity and signaling overhead increase
Solution Approach 1:
Different delta quantization parameters are applied to different sub-partitions based on local characteristics such as residual energy, prediction mode, or block size. This allows each sub-partition to have optimized quantization precision tailored to its specific content, improving overall compression efficiency while managing signaling overhead through selective parameter adjustment.
Data Source
AI summary
Methods, devices, and systems related to video processing are disclosed. In one example aspect, a method of video processing includes performing a conversion between a block of a current picture of a video and a coded representation of the video using an intra-subblock partitioning (ISP) mode. A prediction is determined for each sub-partition using an intra-prediction process based on samples in the current picture using the ISP mode. The block is partitioned into multiple sub-partitions including a first sub-partitions having a same top-left corner position as a top-left corner position of the block.