Vehicle Detection and Tracking for Traffic Monitoring System

This paper presents a real-time video traffic monitoring application based on object detection and tracking, for determining traffic parameters such as vehicle velocity and number of vehicles. In detection step, background modeling approach based on edge information is proposed for separating moving foreground objects from the background. An advantage of edge is more robust to lighting changes in outdoor environments and requires significantly less computing resource. In tracking step, optical flow Lucas-Kanade (Pyramid) is applied to track each segmented object. The proposed system was evaluated on six video sequences recorded in various daytime environment