Abstract
In this study, we focus on improving the performance of raindrop removal. Conventional methods for still images and videos have limited available information, leading to performance constraints. Therefore, we propose a novel approach focusing on camera placement in autonomous driving, utilizing images from other cameras captured at the same time. In the proposed method, the camera images without raindrops are used for raindrop removal by transforming the viewpoint. The proposed method significantly improves the baseline method in terms of subjective evaluation and object detection, and the effectiveness of the proposed method is demonstrated.
