A core aspect of advanced driver assistance systems (ADAS) is to support the driver with information about the current environmental situation of the vehicle. Bad weather conditions such as rain lead the robustness of most surveillance and driver assistance system and might occlude regions of the windshield or a camera lens and therefore affect the visual perception. Hence, the automated detection of raindrops has a significant importance for video-based ADAS.
We perform monocular raindrop detection in single images based on photometric raindrop model. Our method is capable of detecting raindrops precisely, even in front of complex backgrounds. The effectiveness is demonstrated by a significant increase in image registration accuracy which also allows for successful image restoration.
To find features that differentiate raindrops from other occlusions, as well as to identify thick and thin raindrops, we need to analyze the appearance of patches along individual trajectories and the consistency of forwarding/backward motion. For this, we first need to know the image formation model of raindrops, and the computation of long-range trajectories.