DDR: A network of image deraining systems for dark environments

Abstract

In the domain of computer vision, addressing the degradation of image quality under adverse weather conditions remains a significant challenge. To tackle the challenges of image enhancement and deraining in dark settings, we have integrated image enhancement and deraining technologies to develop the DDR (Dark Environment Deraining Network) system. This specialized network is designed to enhance and clarify images in low-light conditions compromised by raindrops. DDR employs a strategic divide-and-conquer approach and an apt network selection to discern patterns of raindrops and background elements within images. It is capable of mitigating noise and blurring induced by raindrops in dark settings, thus enhancing the visual fidelity of images. Through testing on real-world imagery and the Rain LOL dataset, this innovative network offers a robust solution for deraining tasks in dark conditions, inspiring advancements in the performance of computer vision systems under challenging weather scenarios. The research of DDR provides technical and theoretical support for improving image quality in dark environment.