VISION-BASED VEHICLES DETECTION IN COMPLEX TRAFFIC SCENES
Chunbo Wang, Weidong Zhang, Wenyuan Zhang, XU Xiao-ming · 2000
This paper describes a new moving vehicle detection method. The process involves the moving regions first being coarsely detected using the difference of three consecutive frames in an image sequence. The results of motion segmentation are then regulated using the region information that is produced by a graph-based static segmentation method. A template is then presented in order to improve the segmentation results by enhancing the contrast of image regions. This method is shown to effectively locate moving vehicles in a complex background.