Transform Unit Partitioning Reduces HEVC Encoding Complexity

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

High encoding complexity in HEVC due to independent TU and PU partitions, leading to increased processing time and computational complexity, especially in selecting the best merge candidate for 2N×2N CU merge mode.

Innovation Solution

Modifying the TU presentation scheme to relate TU size to PU size, using the split_transform_flag syntax element to indicate 2N×2N or N×N transform unit partitions, and simplifying the R-D optimization process by estimating motion estimation costs for reduced computational complexity.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If TU partition is made independent of PU partition with max RQT depth of 3, then transform processing flexibility is improved, but encoding complexity and processing time increase

Engineering Contradiction:

Improvetransform processing flexibilityVSAvoidencoding complexity

Core Design Contradiction:

Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The patent merges the TU partitioning process with the PU partitioning process by establishing that TU size is determined by PU size. Specifically, for a 2N×2N CU with 2N×2N partition type, the TU size is set to 2N×2N; for other partition types, the TU size is set to N×N. This integration eliminates the need for independent RQT syntax processing and reduces encoding complexity while maintaining transform processing flexibility.

Inventive Principle:

Principle #5Merging (Combining)

2Manufacturing precision

If independent TU partition with max RQT depth of 3 is used, then transform representation is improved, but processing time increases

Engineering Contradiction:

Improvetransform representationVSAvoidprocessing time

Core Design Contradiction:

Manufacturing precisionVSLoss of time

Solution Approach 1:

The patent combines the determination of TU size with the PU partition type decision. By setting TU size based on PU size (2N×2N for 2N×2N partition type, N×N for other types), the system eliminates redundant syntax processing steps and reduces processing time while maintaining effective transform representation.

Inventive Principle:

Principle #5Merging (Combining)

3Productivity

If complex R-D optimization is performed for selecting best merge candidate, then coding efficiency is improved, but computational complexity increases

Engineering Contradiction:

Improvecoding efficiencyVSAvoidcomputational complexity

Core Design Contradiction:

ProductivityVSDevice complexity

Solution Approach 1:

The patent applies partial action by performing R-D optimization only for the 2N×2N CU merge mode rather than all merge modes. The complex R-D cost calculation is selectively applied only when needed (for 2N×2N partition type), while other partition types use simpler processing, thus reducing overall computational complexity while maintaining coding efficiency where most beneficial.

Inventive Principle:

Principle #16Partial or excessive action

Data Source

PatentUS9788019B2Method and apparatus of transform unit partition with reduced complexity

Publication Date: 2017.10.10 HFI INNOVATION INC

AI summary

Transform unit (TU) partition method and apparatus depending on the coding unit (CU) partition and prediction unit (PU) partition are disclosed. In one embodiment, the maximum TU size is restricted to the minimum of PU width and height, except for a 2N×2N coding unit with the 2N×2N partition type. In another embodiment, the maximum TU size equals to maximum of PU width and height, and the minimum TU size equals to minimum of the PU width and height, except for a 2N×2N coding unit with the 2N×2N partition type. In yet another embodiment, the maximum TU size is equal to the maximum of PU width and height, and the minimum TU size is equal to the minimum of PU width and height except for a 2N×2N CU with 2N×2N partition type.