Image restoration method based on two-stage multi-scale generative adversarial network
To solve the problem of insufficient use of image scale information and incorrect reconstruction of glasses structure in face image restoration task,a two-stage multi-scale generative adversarial network restoration model is proposed.In the first stage of the model,U-Net coarse reconstruction network with improved loss is introduced,three different loss functions are fused to improve the reconstruction ability of the generator,double discriminator is used to consider the global information and local information,and a mixed domain attention mechanism is proposed to focus on the spatial and channel information of the image.In the second stage,a new feature enhancement module is constructed to enhance the network's ability to extract details and express structures.The experimental results show that this method can recover all kinds of missing images and effectively restore face images wearing glasses.The peak signal-to-noise ratio,structural similarity and perceived similarity evaluation indexes of the method were improved by 3.81%,2.65%and 0.45%,respectively.
NETL
NSTL
万方数据
