Image Denoising with CNN-Based Attention

Noise removal is one of the most commonly used processes in computer vision. Noise removal improves the quality of the image, thereby improving the performance of computer vision algorithms and providing user pleasing. In this study, we aim to improve the performance of noise removal by adding an efficient attention module, the Convolutional Block Attention Module (CBAM), to the Fast and Flexible Denoising Network (FFDNet) model with an adjustable noise level map as input. By adding the CBAM module to the convolutions used in FFDNet, the CNN's representational power was increased and successful results were obtained. The proposed method achieved high PSNRs in quantitative experiments on different datasets, and in qualitative experiments we observed that the denoised images are close to the target images.