CBAM-DnCNN: An Improved Method For Image Denoising
Guofeng Cai · 2023
In modern society, with the wide application of digital images in the fields of social media, medical diagnosis, security monitoring, and automatic driving, the quality of images has become particularly critical for information transfer and decision-making. With the continuous development of artificial intelligence and computer vision, Image denoising techniques are gradually applied to improve image quality. In this paper, an Image denoising method based on a modified DnCNN network (CBAM-DnCNN) is proposed. Our method first convolves the noisy image to extract features, then captures the spatial structure of the image through the Convolutional Block Attention Module, and then feeds it to this DnCNN network for residual learning. In addition, the combination of BCEWithLogitsLoss and MSELoss is used as a loss function during training and the introduction of these methods can improve the quality of the results. This work provides a new method for image noise reduction and proves the effectiveness of the method in experiments.