Video Coding Block Partitioning for Transform Unit Efficiency
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding systems face inefficiencies in compressing video data due to limitations in transform unit (TU) size, which can leave residual samples untransformed when a coding unit (CU) is larger than the max TU size, resulting in suboptimal compression ratios and image quality.
Innovation Solution
A method that splits a CU into sub-CUs based on comparisons of CU dimensions to max TU dimensions, allowing multiple TUs to be applied, thereby maximizing the transformation of residual samples and improving coding efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a single TU is applied to a CU when max TU size is smaller than CU size, then device complexity is reduced, but coding efficiency deteriorates due to untransformed residual samples
Solution Approach 1:
The patent divides a large CU into multiple sub-CUs when the CU size exceeds the maximum TU size. This segmentation allows multiple TUs to be applied to different sub-CUs, ensuring complete transformation of all residual samples while maintaining manageable processing complexity for each individual TU application.
2Productivity
If multiple TUs are applied to a CU to maximize residual transformation, then coding efficiency is improved, but device complexity increases
Solution Approach 1:
The patent segments the CU into sub-CUs based on the maximum TU size constraint, allowing systematic application of multiple TUs. Each sub-CU receives at most one TU, simplifying the decision logic to a straightforward segmentation and mapping process rather than complex multi-TU placement optimization.
Solution Approach 2:
The patent changes the parameter of TU application from a single-unit approach to a multi-unit approach by introducing the segmentation parameter. The decision to apply one or multiple TUs is based on comparing CU dimensions with maximum TU dimensions, transforming a complex optimization problem into a simple parameter-based decision rule.
3Device complexity
If CU is not split when max TU size is smaller than CU size, then device complexity is reduced, but compression ratio deteriorates due to untransformed residual samples
Solution Approach 1:
The patent implements segmentation of CU into sub-CUs only when necessary (when CU size exceeds maximum TU size). This conditional segmentation ensures that all residual samples are transformed when needed, maximizing compression ratio without adding unnecessary partitioning complexity for cases where a single TU suffices.
4Productivity
If max TU size is increased to match CU size, then coding efficiency is improved by transforming all residual samples, but adaptability deteriorates for smaller CUs
Solution Approach 1:
The patent makes the TU application dynamic by allowing the system to adapt between single-TU and multi-TU modes based on the relative sizes of CU and maximum TU. This dynamic behavior enables optimal performance across varying CU sizes while maintaining a fixed maximum TU size constraint, providing both high coding efficiency and broad adaptability.
Data Source
AI summary
A video coding mechanism is disclosed. The mechanism includes selecting a split mechanism to split a coding unit (CU) into sub-CUs for application of one or more transform units (TUs), the selection of the split mechanism based on comparing a CU width to a max TU width and comparing a CU height to a max TU height. The selected split mechanism is applied to the CU to obtain sub-CUs. A residual of one of the sub-CUs is determined. The residual includes a difference between sample values for the sub-CU and prediction samples for the sub-CU. The TUs are applied to transform the residual of the CU based on results of the selected split mechanism. A transformed residual for the CU is encoded into a bitstream.