Deep learning has been widely applied in many computer vision tasks due to its impressive capability of automaticfeature extraction and classi cation. Recently, deep neural networks have been used in image denosing, but mostof the proposed approaches were designed for Gaussian noise suppression. Therefore, in this paper, we address theproblem of impulsive noise reduction in color images using Denoising Convolutional Neural Networks (DnCNN).This network architecture utilizes the concept of deep residual learning and is trained to learn the residual imageinstead of the directly denoised one. Our preliminary results show that direct application of DnCNN allows toachieve signi cantly better results than the state-of-the-art lters designed for impulsive noise in color images.
展开▼