Zero-DCE with HSV loss for Low-Light Image Enhancement

Many surveillance or reconnaissance images are taken at night without auxiliary light. The images are usually very dim and need to be enhanced. Zero-Reference Deep Curve Estimation (Zero-DCE), which formulates light enhancement as a task of image-specific curve estimation with a deep network. However, Zero-DCE can enhance the images and perform well, color compensation is required. The RGB and HSV color space are compared and analyzed, and the differences, advantages and disadvantages between the two are compared. The HVS loss algorithm is considered to be converted from RGB to HSV color field, and the HSV algorithm is constructed in order to obtain new effects in color compensation. Extensive experiments on various benchmarks demonstrate the advantages of our method over state-of-the-art methods qualitatively and quantitatively.