DA-Zero- DCE

Tunnel images are affected by the shooting environment, resulting in uneven illumination distribution, local occlusion, and a high amount of noise. To address the overexposure and distortion issues in existing image enhancement algorithms, this paper proposes a tunnel image enhancement algorithm called DA-Zero-DCE (Denoising-Attention based Zero-Reference Deep Curve Estimation). Firstly, based on the Zero-DCE model, the U-Net is used to improve the backbone network DCE-Net for curve estimation, and a coordinate attention mechanism is added to enhance the dark light perception ability of local image areas. Secondly, the NAF-Net noise removal module is added after the curve estimation backbone network to effectively suppress the noise after low-light enhancement by Zero-DCE. In addition, to alleviate the distortion and overexposure of the enhanced images, the 4-neighborhood calculation method of the spatial consistency loss function is extended to an 8- neighbor hood calculation method, enhancing the smoothness of the output results. Through the ablation experiment conclusions on the LOL dataset, the DA-Zero-DCE model compared to the Zero-DCE model has a PSNR improvement of about 10dB and an SSIM improvement of about 0.1, verifying the effectiveness and feasibility of the model.