Image restoration method and system

By extracting information from the three dimensions of image—channel, space, and frequency—and combining an encoder-decoder architecture with multiple attention modules, this technology solves the problems of insufficient sensitivity to texture details and loss of frequency information in existing image restoration techniques, achieving highly efficient image restoration results.

CN121998874APending Publication Date: 2026-05-08SICHUAN UNIV
2 Cites 0 Cited by

Patent Information

Application Number
CN202610468847.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-04-10
Publication Date
2026-05-08

AI Technical Summary

Technical Problem

Existing technologies are not sensitive enough to local texture details in image restoration, suffer from severe loss of frequency information, and have high computational complexity, making them difficult to adapt to complex degradation scenarios.

Method used

Adopting the three-domain divide-and-conquer approach, information is comprehensively extracted from the three dimensions of image: channel, space, and frequency. Hierarchical representations are learned through an encoder-decoder architecture, and skip connections are used to supplement the original features. Image inpainting is performed by combining transposed channel attention, spatial awareness, and frequency learning modules.

Benefits of technology

While reducing computational complexity, it restores texture details in complex and degraded scenes, enhances the network's ability to represent various features, and generates more natural images.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The invention relates to the field of image processing, in particular to an image restoration method and system, and the method comprises the steps: inputting a degraded image into an image restoration network; the method comprises the following steps: respectively extracting channel information, space information and frequency domain information of a degraded image from three dimensions of an image channel, space and frequency through an image restoration network, restoring the degraded image by utilizing the extracted channel information, space information and frequency domain information, and outputting a restored image. While degeneration is removed, texture details in multiple complex degeneration scenes can be recovered, and the calculation complexity of the model can be reduced.
Need to check novelty before this filing date? Find Prior Art