An Image Denoising Algorithm Based on Multi-scale Residual Channel Attention Network

Aiming at the problem of poor denoising effect of existing algorithms on complex color texture images, an image denoising algorithm based on multi-scale residual channel attention is proposed. Firstly, the shallow features of the image are extracted; then these features are fed into multiple multi-scale residual attention feature extraction modules to extract the depth features of the image at multiple scales; the depth features are fused using the non-local-sparse local fusion module; finally, the denoised image is reconstructed through the reconstruction layer to generate a high-quality denoised image. The experimental results show that compared with MPRNet, MIRNet, Uformer, ADNet, DudeNet and other methods, this method has better denoising performance on images with complex colors and textures, as well as better adaptability to the diversity of noise in complex scenes.