An effective shadow detection approach in traffic scenes

Accurate moving objects segmentation is an essential problem in intelligent video surveillance system. However, the existence of unexpected moving cast shadows frequently lead to errors in further scene analysis. This paper presents a novel method that combines color space and corner feature to detect and remove cast shadows of moving vehicles in traffic scenes. The two features cooperate well to distinguish vehicles from cast shadows by proposed algorithm, Multiple Masks Method. The proposed algorithm has been tested on sufficient video sequences taken under different illumination conditions, various shadow orientations and shadow sizes. The results have revealed that shadows can be successfully eliminated and thus good vehicle segmentation can be obtained.