Light- weight Zero-Reference-based Image Enhancement for Low-Light Images

Images captured under low-light conditions often suffer from overall darkness and poor visual quality. In this study, we propose a light-weight zero-reference image enhancement method to improve the visual quality of low-light images. The proposed method is composed of some parts: image partition, transformation map estimation, and pixel value adjustment. To achieve transformation map estimation, a light-weight network, CSPDCE-Net, is developed. For performance evaluation, two image datasets, SICE and LOL, are selected here. For visual quality, the average PSNR and SSIM of the proposed method can achieve 17.01dB and 0.71, respectively. Compared with the existing methods, the proposed method can achieve better performance in terms of PSNR and SSIM. The experimental results show that the proposed method not only can process low-light images well but also is superior to some existing methods.