Vehicle Segmentation from Video Using Color and Edges

Intelligent transportation systems or ITS play a very important role in traffic management. Traffic congestion is widespread in big cities. The key success factor of traffic management is accurate traffic data collection. Traffic video processing is a widely used technique in traffic data collection. However, the changes in environment where traffic video is collected can adversely affect the reliability of the traffic video processing results. In this paper, a robust vehicle segmentation approach is presented. An improvement to color-based statistical background subtraction is proposed. It is shown that improvement in reliability and robustness to lighting changes in different camera views can be achieved when compared to results from conventional techniques.