A method to segment moving vehicle cast shadow based on wavelet transform

In many image analysis and interpretation applications, shadows interfere with fundamental tasks such as object extraction and description. According to illumination, shadows interfere with moving vehicle extraction and location and recognition. For this reason, shadow segmentation is an important step in real-time vehicle recognition system. In this paper, we propose a simple and effective method for detection of moving cast shadows on a traffic surveillance scene. The proposed method exploits spectral and geometrical properties of shadows and relationship between the point in shadow region and space position and vehicle shape. Firstly, the cast shadows can be rough detected by spectral properties, and then feature points of occluding function are detected using wave transform, finally, the boundary between self-shadow and cast shadow is detected. The proposed method does not know in advance the light source direction and the color information of vehicle and background texture information. Our experimental results demonstrate that the proposed cast shadows segmentation method can detect the shadows regions accurately and completely. This is the foundation for future vehicle recognition and understanding.