Image restoration method based on similar convex set projection algorithm

Abstract

The invention discloses an image restoration algorithm based on a similar convex set projection algorithm, which is used for solving the problem that a large amount of noise exists in an existing image restoration network feature map. The method comprises the following steps: 1) constructing a training sample set and a test sample set; 2) constructing a displayed feature denoising module; 3) constructing an image restoration network model PL-DSR based on a similar convex set projection algorithm; 4) training the PL-DSR network model; 5) testing the PL-DSR network model; 6) generating a restored image. A displayed feature denoising module is constructed on the basis of a traditional convex set projection algorithm and used for suppressing noise in a feature map and introducing the noise into a classical EDSR network, and an image restoration network PL-DSR based on a similar convex set projection algorithm is constructed. The network can effectively suppress noise in the feature map, extract structured clearer features, further improve the recovery effect and the visual sensory experience of people while reducing the complexity of the model, and can be used for completing later manufacturing and processing of photography and movie and television works.

Description

technical field [0001] The invention belongs to the technical field of digital image processing, and specially designs an image restoration method based on a similar convex set projection algorithm to restore degraded images, which can be used to complete the post-production processing of photography and film and television works. Background technique [0002] An important branch in the field of computer vision is image generation, which includes image restoration, image coloring, image semantic segmentation, image or video style conversion, etc. In the field of computer vision, image restoration is an ill-conditioned inverse problem, that is, in the case of specifying a low-resolution image, the prior knowledge of the degradation process is used to establish an image degradation model and restore the original high-resolution image. [0003] Traditional image restoration models mainly include: 1) algorithms based on interpolation (nearest neighbor method, bilinear interpolat...

AI Technical Summary

Problems solved by technology

However, none of the existing image restoration networks considers the influence of the degree of noise in the features on the restoration results

Images