TSN-CA: A Two-Stage Network with Channel Attention for Low-Light Image Enhancement

Xinxu Wei, Xian-Shi Zhang, Yongjie Li

Abstract: Low-light image enhancement is a challenging low-level computer vision task because after enhancing the brightness of the image, amplified noise, color distortion, detail loss, blurred edges, shadow blocks and halo artifacts occur. In this paper, we propose a Two-Stage Network with Channel Attention (denoted as TSN-CA) to enhance the brightness of the low-light image and restore the enhanced images from various kinds of degradation. In the first stage, the brightness of the low-light image is enhanced in HSV space using a U-Net based enhancement network, which uses the information of H and S channels to help recover the details in V channel. In the second stage, Channel Attention (CA) mechanism is integrated into the skip connection of a U-Net in order to restore the brightness-enhanced image from several kinds of degradation in RGB space, especially the halo and shadow blocks. Extensive experiments performance demonstrate that our method achieves excellent effect on brightness enhancement as well as denoising, detail preservation and halo artifact elimination.