Video Coding In-Loop Filtering Parameter Signaling

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

In video coding standards like VVC, the determination of in-loop filtering parameters is delayed due to their dependency on other video processing procedures, leading to a deterioration in video encoding and decoding performance.

Innovation Solution

The method involves signaling and parsing CTU coded blocks data and CTU in-loop filtering parameters in distinct Network Abstract Layer (NAL) units, allowing for independent processing and parallelization of these parameters with the coded block data, reducing dependencies and enabling earlier start of in-loop filtering.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If in-loop filtering parameters are determined after other video processing procedures, then the filtering can be applied correctly, but the encoding and decoding performance deteriorates due to delayed processing

Engineering Contradiction:
Improvefiltering correctnessVSAvoidencoding and decoding performance
Core Design Contradiction:
ReliabilityVSProductivity

Solution Approach 1:

The patent segments the bitstream into distinct NAL units: one for coded block data and another for in-loop filtering parameters. This segmentation allows the filtering parameters to be processed independently and in parallel with the coded block data, eliminating the sequential dependency that caused performance deterioration while ensuring the filtering is still applied correctly to the appropriate blocks.

Inventive Principle:
Principle #1Segmentation

2Stability of the object's composition

If in-loop filtering parameters are processed sequentially after coded block data, then processing order is maintained, but decoding time increases and real-time processing at high resolutions becomes difficult

Engineering Contradiction:
Improveprocessing orderVSAvoiddecoding time
Core Design Contradiction:
Stability of the object's compositionVSLoss of time

Solution Approach 1:

The patent enables preliminary processing of in-loop filtering parameters by placing them in a separate NAL unit that can be parsed and prepared in advance, independent of the coded block data processing. This preliminary action allows the filtering parameters to be ready before they are needed, reducing the critical path duration and enabling real-time processing at higher resolutions and bitrates.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP4475529A1Video processing method
Publication Date: 2024.12.11 BEIJING XIAOMI MOBILE SOFTWARE CO LTD
  • EP4475529A1 patent drawingFigure 1~2
  • EP4475529A1 patent drawingFigure 3~4
  • EP4475529A1 patent drawingFigure 5

AI summary

The present disclosure relates to signaling in a bitstream coded block data and in-loop filtering parameters of a video sequence comprising video pictures, each video picture being split into multiple coding tree units, denoted CTU, each CTU comprising CTU blocks, CTU coded blocks data being obtained for each CTU by encoding blocks of said CTU, reconstructed CTU blocks associated with a CTU being obtained by decoding CTU coded blocks data associated with said CTU, reconstructed CTU blocks associated with a CTU being filtered using in-loop filtering parameters associated with said CTU, denoted CTU in-loop filtering parameters, wherein CTU coded blocks data associated with a CTU and CTU in-loop filtering parameters associated with said CTU are signaled in distinct Network Abstract Layer units, called NAL units.