A Novel Conditional Generative Adversarial Network Based On Graph Attention Network For Moving Image Denoising
Weihong Shen · DOAJ (DOAJ: Directory of Open Access Journals) · 2022
Aiming at the problem that the noise in the process of image acquisition and transmission leads to the degradation of the subsequent image processing capability, we propose a conditional generative adversarial network (CGAN) based on graph attention network for moving image denoising. The CGAN can effectively extract depth features and avoid losing details. The discriminant network is constructed based on fully convolutional network, so pixel classification can be obtained to improve the accuracy of discrimination. In addition, in order to improve the denoising ability and preserve the image details as much as possible, a graph attention network is proposed. The compound loss function is constructed based on confrontation loss, visual perception loss and mean square error loss. Finally, the adaptive weighted average is used to fuse the three-channel output information to obtain the final denoised image. Experimental results show that compared with other state-of-the-art denoising algorithms, the proposed algorithm can effectively remove image noise and restore original image details.