An algorithm of traffic parameters extraction for multi-lane roads

A novel method for extracting traffic flow parameters for multi-lane roads is proposed. The presented method based on the technology of virtual loops. Firstly, complete of target vehicle information are extracted from traffic video sequences by using the background subtraction technique and shadow model in color space. Secondly, vehicles are detected through the virtual loops in gray space. Meanwhile, vehicle speed and vehicle type are extracted by using the projection technique. Compared with traditional methods, this method focus on the traffic flow parameters extraction for multi-lane roads and solves the vehicle lane-changing problem. Experimental results of several real traffic video sequences demonstrate robustness and effectiveness of the presented method.