Adaptive Loop Filtering by Temporal Layer for Lower Video Decoding Load

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing block-based hybrid video coding systems face high computational complexity due to the extensive use of adaptive loop filters (ALFs), particularly in block classification and filtering processes, which hinder efficient compression and decoding of high-resolution video signals.

Innovation Solution

The proposed method reduces computational complexity by selectively applying adaptive loop filtering (ALF) procedures based on temporal layer and gradient calculation, using subsampling techniques for gradient computation and disabling ALF for higher temporal layers, and employing frame-adaptive ALF skip to minimize redundant processing.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Manufacturing precision

If adaptive loop filtering is applied to all blocks in video coding, then video quality is improved, but computational complexity increases significantly

Engineering Contradiction:
Improvevideo qualityVSAvoidcomputational complexity
Core Design Contradiction:
Manufacturing precisionVSDevice complexity

Solution Approach 1:

The patent applies ALF selectively to specific blocks based on their characteristics rather than uniformly to all blocks. Block classification identifies regions that benefit from filtering, applying ALF only where needed to maintain video quality while reducing overall computational complexity.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The patent uses subsampling techniques to compute gradients at reduced resolution, performing partial gradient calculations that are sufficient for block classification decisions. This partial action reduces the number of computations required while maintaining adequate filtering performance.

Inventive Principle:
Principle #16Partial or excessive action

2Measurement precision

If gradient calculation is performed for all pixels in block classification, then filtering accuracy is improved, but processing time increases

Engineering Contradiction:
Improvefiltering accuracyVSAvoidprocessing time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent calculates gradients only for a subset of pixels rather than all pixels in the block. This partial gradient calculation provides sufficient accuracy for block classification while significantly reducing the number of operations and processing time required.

Inventive Principle:
Principle #16Partial or excessive action

Solution Approach 2:

The patent divides the block processing into segments, applying gradient calculation only to selected regions or pixel subsets. This segmentation approach maintains filtering accuracy for critical areas while avoiding redundant calculations in other regions.

Inventive Principle:
Principle #1Segmentation

3Manufacturing precision

If ALF is applied to all temporal layers, then video quality is maintained, but redundant processing occurs in higher temporal layers

Engineering Contradiction:
Improvevideo qualityVSAvoidencoding efficiency
Core Design Contradiction:
Manufacturing precisionVSProductivity

Solution Approach 1:

The patent dynamically adjusts ALF application based on temporal layer characteristics. Higher temporal layers with motion-compensated prediction are identified as having reduced benefit from ALF, so filtering is adaptively disabled for these layers, eliminating redundant processing while maintaining quality where it matters.

Inventive Principle:
Principle #15Dynamics

Data Source

PatentEP3704858B1Methods for simplifying adaptive loop filter in video coding
Publication Date: 2025.12.03 INTERDIGITAL VC HOLDINGS INC
  • EP3704858B1 patent drawingFigure 1
  • EP3704858B1 patent drawingFigure 2
  • EP3704858B1 patent drawingFigure 3(a)~3(c)

AI summary

Systems, methods and instrumentalities are disclosed for adaptively selecting an adaptive loop filter (ALF) procedure for a frame based on which temporal layer the frame is in. ALF procedures may vary in computational complexity. One or more frames including the current frame may be in a temporal layer of a coding scheme. The decoder may determine the current frame's temporal layer level within the coding scheme. The decoder may select an ALF procedure based on the current frame's temporal layer level. If the current frame's temporal layer level is higher within the coding scheme than some other temporal layer levels, an ALF procedure that is less computationally complex may be selected for the current frame. Then the decoder may perform the selected ALF procedure on the current frame.