Rain Streaks Removal in Single Low-light Image
Tong Liu, Zehao Wu, Haowei Li, Wenqi Fu · 2022 41st Chinese Control Conference (CCC) · 2022
Unmanned platforms have been used more and more in our life. Most unmanned platforms obtain environmental information through cameras for subsequent operations. But when working outdoors, unmanned platforms may encounter rainy and low-light environments. Rain streaks and low-light environment will introduce undesired interference and influence the visibility of the images, which affect the safe work of unmanned platforms. To solve this problem, there are several methods proposed to remove rain streaks. However, the existing methods only consider the daytime images, which do not achieve the rain streaks removal in low-light environment. In this paper, we propose a novel approach to achieve rain streaks removal in low-light environment. We firstly decompose the low-light rainy day image into the background layer and the rain layer by using layer priors. And we enhance the background layer by using improved low-light enhancement algorithm, regarding the enhanced background layer as the final result to achieve the enhancement of low-light rainy day images. The image enhanced by our method can recover detailed information well, and has a high degree of naturalness, which provides strong support for the subsequent operations of unmanned platforms. In the end, we proved the superiority of our algorithm through some experiments.