Real Image Blind Denoising Based on Attentional Mechanism and Amplified Receptive Field
DU Yanmo
SHEN Sanmin
ZHANG Bingwei
Abstract：In order to achieve a good balance between spatial information and contextual semantic information in image denois-ing task,a real image denoising network based on attention mechanism and amplified receptive field is proposed.The network is composed of noise estimation and non-blind denoising.The receptive field is expanded by kernel dilation,and the attention mecha-nism is embedded in the forward connection and the horizontal skip connection to capture task-related feature information efficient-ly,and the contextual semantic information is retained to a certain extent.The linear combination of asymmetric loss and total varia-tion loss is used as the loss function to deal with the complex and variable noise types in real images and improve the generalization ability of the model.The experimental results show that the SSIM of the proposed method on SIDD and DND data sets reaches 88%and 95%,and the PSNR reaches 35.48 dB and 39.16 dB.Therefore,it has a good image denoising effect.
Keywords：image denoisingattention mechanismdilated convolutiondeep learning
Publication Date：2025-07-20
Online Publishing Date：2025-09-25(First online date of this platform, not the publication date of the document)
Pages：5( 32-35,85 )