Low-Complexity CIIP and GEO Transform Sets for LFNST and NSPT

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing video coding techniques face challenges in efficiently determining transform sets for combined inter-intra prediction (CIIP) and geometric partitioning (GEO) modes, leading to increased complexity in video encoding and decoding processes.

Innovation Solution

The proposed solution involves deriving additional transform sets for CIIP and GEO modes, allowing video encoders and decoders to signal these sets in the encoded video bitstream, and using decoder-side intra mode derivation processes to determine transform sets based on prediction block samples, thereby reducing the complexity of transform selection.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If multiple transform sets are used for CIIP and GEO modes, then video coding flexibility and adaptability are improved, but the complexity of determining and signaling transform sets increases

Engineering Contradiction:

Improvetransform set selection flexibilityVSAvoidtransform determination complexity

Core Design Contradiction:

Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The patent segments the transform set determination process into distinct phases: first determining a candidate transform set based on prediction mode, then selectively deriving additional transform sets only when needed (e.g., for CIIP and GEO modes). This segmentation reduces overall complexity by avoiding unnecessary transform set derivations while maintaining flexibility when required.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent applies preliminary action by pre-establishing a first set of transform sets that can be directly selected based on prediction mode, and pre-defining the conditions under which additional transform sets are derived. This preliminary structuring simplifies the decision-making process during actual video coding operations.

Inventive Principle:

Principle #10Preliminary action

2Productivity

If additional transform sets are derived for CIIP and GEO modes, then video encoding efficiency is improved, but processing time and computational resources increase

Engineering Contradiction:

Improvevideo encoding efficiencyVSAvoidtransform determination time

Core Design Contradiction:

ProductivityVSLoss of time

Solution Approach 1:

The patent applies partial action by deriving additional transform sets only for specific modes (CIIP and GEO) where they provide benefit, rather than deriving all possible transform sets for all modes. This selective approach maintains encoding efficiency improvements while minimizing unnecessary computational overhead.

Inventive Principle:

Principle #16Partial or excessive action

Solution Approach 2:

The patent implements dynamic transform set determination where the set of available transforms is adaptively adjusted based on the specific prediction mode and coding conditions. This dynamic approach allows the system to optimize between speed and quality on a per-block basis, reducing overall processing time compared to static comprehensive transform set derivation.

Inventive Principle:

Principle #15Dynamics

Data Source

PatentUS20250324053A1Inter CIIP and inter GEO mode transform set selection for lfnst and NSPT in video coding

Publication Date: 2025.10.16 QUALCOMM INC

AI summary

A video coder may determine transform sets for blocks coded with combined inter-intra prediction (CIIP) mode and/or a geometric partitioning (GEO) inter mode. The video coder may determine the transform sets from intra prediction modes derived for the blocks coded with CIIP mode and GEO mode.