Application of traffic conflict technique for traffic safety evaluation at intersection based on image processing

Traffic accidents occur frequently at intersections area because of conflicts among vehicles. Many researchers have been assessing traffic safety, however most of them are based on accident data that happened and caused injury as well as economic damage. Recently, conflict techniques provide general views and solutions to prevent early collisions. This paper proposes a method of conflict analysis using image processing technique and fuzzy comprehensive evaluation. Based on vehicles detection, the conflict parameters collected from two intersections in Hanoi, Vietnam were processed and evaluated by fuzzy comprehensive evaluation to give out the safety level of conflict points. The experimental results show that the proposed methods have successfully shown the distribution and location of dangerous conflict points according to the actual situation of the two intersections. Based on our results, authorities can consider reorganizing traffic light pattern at these intersections to reduce possible collisions.