Benchmark for the robustness of image features in rainy conditions

Computer vision systems are increasingly present on roadways, both on the roadside and on board vehicles. Image features are an essential building block for computer vision algorithms in a road environment. Eight of the most representative image features in a road environment are selected on the basis of a literature review, and their robustness in rainy conditions is evaluated. In order to do this, a new evaluation method is proposed, which is then applied to a new weather-image database. This database contains rain typical of latitudes with a temperate climate (\(0{-}30\, \mathrm{mm}\,\mathrm{h}^{-1} \)), various camera settings and images with natural rain and images with digitally simulated rain. Image features based on pixel intensity and those that use vertical edges are sensitive to rainy conditions. Conversely, the Harris feature and features that combine different edge orientations remain robust for rainfall rates of \(0{-}30\, \mathrm{mm}\,\mathrm{h}^{-1} \). The robustness of image features in rainy conditions decreases as the rainfall rate increases. Finally, the image features most sensitive to rain have potential for use in a camera-based rain classification application.