Template Analysis for Low-Signaling Transform Selection in Video Coding

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing video coding systems face challenges in efficiently selecting transform sets without incurring high signaling costs, which affects the compression efficiency of digital video signals.

Innovation Solution

Adaptive template-based multiple transform set selection (MTSS) that allows switching between transform sets based on template analysis, determining the highest ranked candidate transform set without explicit signaling, using measures such as energy sums or distortion metrics.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If explicit signaling is used to identify the transform set, then the transform set selection accuracy is improved, but the signaling cost increases

Engineering Contradiction:
Improvetransform set selection accuracyVSAvoidsignaling overhead
Core Design Contradiction:
Measurement precisionVSLoss of information

Solution Approach 1:

The decoder autonomously determines the optimal transform set by performing template analysis and calculating matching metrics without receiving explicit transform set indications from the bitstream. The encoder and decoder both independently select the same transform set based on identical analysis, eliminating the need for signaling.

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The system changes the selection criterion from explicit signaling to template matching metrics. By analyzing the correlation between prediction templates and residual templates, the system dynamically selects transform sets based on content characteristics, achieving accurate selection without additional signaling bits.

Inventive Principle:
Principle #35Parameter changes

2Productivity

If multiple transform sets are evaluated, then the compression efficiency is improved, but the computational complexity increases

Engineering Contradiction:
Improvecompression efficiencyVSAvoidcomputational complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

Instead of exhaustively evaluating all possible transform sets, the system evaluates only the necessary candidate transform sets (e.g., DCT2 and DST7) based on template analysis results. This partial evaluation approach achieves sufficient compression efficiency while avoiding excessive computational complexity.

Inventive Principle:
Principle #16Partial or excessive action

3Adaptability or versatility

If template analysis is performed for transform set selection, then the adaptability is improved, but the processing time increases

Engineering Contradiction:
Improvetransform set adaptabilityVSAvoidprocessing time
Core Design Contradiction:
Adaptability or versatilityVSLoss of time

Solution Approach 1:

The template analysis is performed during the encoding process to pre-determine the optimal transform set. Both encoder and decoder perform this analysis beforehand, allowing the transform set to be selected adaptively without adding processing time during decoding or transmission.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP4633152A1Template based multiple transform set selection
Publication Date: 2025.10.15 INTERDIGITAL CE PATENT HOLDINGS SAS
  • EP4633152A1 patent drawingFigure 1A
  • EP4633152A1 patent drawingFigure 1B
  • EP4633152A1 patent drawingFigure 1C

AI summary

Systems, methods, and instrumentalities are disclosed for performing template based multiple transform set selection (MTSS). A device, such as a video decoding device, may perform (e.g., be configured to perform) one or more of the following actions. The device may determine a prediction block associated with a video block. The device may obtain a residual block associated with the video block, for example, based on the prediction block and the video block. The device may select a transform set for the video block from candidate transform sets based on respective measures associated with the candidate transform sets. The respective measure for each of the candidate transform sets may be determined based on respective transform coefficients of the candidate transform set. The device may perform a transform for the video block based on the selected transform set.