NNPF SEI Signaling for Video Bitstream Quality Enhancement

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing video coding standards lack efficient mechanisms for signaling and utilizing neural-network post-filters (NNPFs) in video bitstreams, limiting the ability to enhance video quality through advanced processing techniques.

Innovation Solution

The implementation of a neural-network post-filter (NNPF) SEI message for video bitstreams, which includes signaling the usage and characteristics of NNPFs, allowing for enhanced video quality improvements such as upsampling, colorization, and bit depth enhancement.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Manufacturing precision

If existing video coding standards are used without NNPF signaling mechanisms, then the bitstream structure remains simple and compatible, but video quality enhancement capabilities are limited

Engineering Contradiction:
Improvevideo qualityVSAvoidbitstream structure
Core Design Contradiction:
Manufacturing precisionVSDevice complexity

Solution Approach 1:

The patent segments the bitstream structure by introducing separate SEI messages for NNPF characteristics and NNPF activation. The NNPF characteristics SEI message carries filter parameters and properties, while the NNPF activation SEI message controls when the filter is applied. This segmentation allows independent optimization of video quality enhancement without complicating the core video coding structure.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent introduces SEI messages as intermediary structures between the video bitstream and the NNPF processing. These messages act as carriers that transport NNPF configuration data and activation commands without directly modifying the video bitstream format, thereby maintaining compatibility while enabling advanced processing capabilities.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Adaptability or versatility

If NNPF characteristics are fully signaled in SEI messages, then video processing flexibility is improved, but message size and processing overhead increase

Engineering Contradiction:
Improvevideo processing flexibilityVSAvoidmessage size
Core Design Contradiction:
Adaptability or versatilityVSQuantity of substance

Solution Approach 1:

The patent applies local quality by signaling only the necessary NNPF characteristics in the SEI messages rather than complete filter configurations. The characteristics message includes selective parameters such as filter type, resolution scaling factors, and color space transformations, allowing flexible video processing while keeping message sizes manageable by including only locally relevant information.

Inventive Principle:
Principle #3Local quality

3Ease of operation

If NNPF is integrated into existing video coding standards, then implementation simplicity is maintained, but advanced video quality enhancement capabilities are lost

Engineering Contradiction:
Improveimplementation simplicityVSAvoidvideo quality enhancement
Core Design Contradiction:
Ease of operationVSManufacturing precision

Solution Approach 1:

The patent achieves universality by designing the NNPF SEI message structure to be compatible with existing video coding standards while supporting multiple video quality enhancement functions. The same SEI message framework can carry different types of NNPF characteristics and activation commands, enabling various advanced processing capabilities (upsampling, colorization, bit depth enhancement) without requiring separate integration mechanisms for each function.

Inventive Principle:
Principle #6Universality (Multi-functionality)

Data Source

PatentUS20250384589A1Neural-network post filter characteristics representation
Publication Date: 2025.12.18 DOUYIN VISION CO LTD
  • US20250384589A1 patent drawing
  • US20250384589A1 patent drawing
  • US20250384589A1 patent drawing

AI summary

A mechanism for processing video data is disclosed. The mechanism includes determining a supplemental enhancement information (SEI) message contains a non-binary syntax element indicating usage of a neural-network post-filter (NNPF). A conversion can then be performed between a visual media data and a bitstream based on the NNPF and the non-binary syntax element.