Low-light Image Enhancement Using Cone-Rod Coupling

Abstract

In low-light environments, the captured image typically exhibits problems such as insufficient brightness, intense noise, and low color contrast, leading to performance degradation in computer vision applications. In this study, we introduce a novel low-light image enhancement method based on the cone-rod coupling mechanism in the human visual system. Specifically, based on the properties of cone and rod cells, two lightweight sub-nets are designed to conduct the tasks of brightness adjustment and color correction with deep gamma correction, respectively. To further enhance the details and filter the noise, an efficient Transformer network is constructed for post-processing. In the experiments, we compare the proposed method with other state-of-the-art techniques through quantitative and qualitative evaluation. The experimental results demonstrate that the proposed method can significantly outperform other methods with more natural color fidelity and balanced brightness distribution.Clinical Relevance-This introduces a bio-inspired low-light image enhancement based on the cone-rod coupling mechanism in the visual system.