Image Bitstream Encoding With Targeted Neural Post-Filter Activation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The increasing demand for high-resolution and high-quality images leads to a surge in transmission and storage costs due to the increased amount of information, necessitating high-efficient image compression technologies.
Innovation Solution
Implementing an image encoding/decoding method that includes neural-network post-filter characteristics (NNPFC) and activation (NNPFA) supplemental enhancement information (SEI) messages to determine and apply appropriate neural-network post-filters (NNPFs) to current pictures, with corresponding SEI messages encoding whether a target NNPF is a base or another NNPF.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If high-resolution and high-quality images are transmitted, then image quality is improved, but transmission cost and storage cost increase
Solution Approach 1:
The patent extracts and transmits only the essential image information through efficient compression algorithms, separating critical visual data from redundant information. This allows high-quality image reconstruction with reduced bitstream size, directly addressing the contradiction between image quality and transmission volume.
Solution Approach 2:
The patent employs advanced compression parameters and neural network-based post-filters that dynamically adjust encoding parameters to maintain perceptual image quality while significantly reducing the amount of transmitted data. The neural networks optimize the balance between compression ratio and visual fidelity.
2Productivity
If neural-network post-filter is applied to improve image quality, then decoding efficiency is improved, but device complexity increases
Solution Approach 1:
The patent segments the image decoding process into distinct stages, with the neural-network post-filter operating as a separate module after basic decoding. This modular approach allows the complex neural network to be independently optimized and managed, reducing overall system complexity while maintaining decoding efficiency.
Solution Approach 2:
The patent performs preliminary decoding operations before applying the neural-network post-filter, preparing the image data in advance to optimize neural network processing. This staged approach simplifies the neural network's computational burden and improves overall decoding efficiency.
Data Source
AI summary
Disclosed herein are an image decoding method including obtaining a neural-network post-filter characteristics (NNPFC) supplemental enhancement information (SEI) message and a neural-network post-filter activation (NNPFA) SEI message, determining neural-network to be used as a neural-network post-filter (NNPF) based on the NNPFC SEI message, determining whether activating a target NNPF to be applied to a current picture or not based on the NNPFA SEI message, wherein the NNPFA SEI message includes a NNPFA target flag indicating whether the target NNPF is a base NNPF or another NNPF, and wherein a NNPFA target pictures for which the target NNPF is activated by the NNPFA SEI message is included in a NNPFC target pictures, to which the NNPFC SEI message corresponding to the target NNPF indicated by the NNPFA target flag pertains.