The over-height vehicle detection using the computer vision method

Abstract

Over-height vehicles are not only threats for the over-height vehicles themselves, but they also bring about a hazard to the tunnels they pass in, and to other drivers on the road way when a clash happens with a low clearance structure. Therefore, it is needed an over-height detection system which is affordable yet also reliable. Nowadays, there are the over-height detection system using laser and infrared. However, they are quite pricy. In this research a computer vision-based system is proposed to detect the height of vehicles and provide a warning for the over-height vehicles. The height determination was conducted using Gaussian Mixture Model (GMM) and blob detection. The GMM is used to detect the vehicle and the blob detection is applied to produce the vehicle coordinates which determines the direction of movement and the over-height vehicle. The over-height vehicles are detected by using 5 (five) variance total frames within 3 (three) conditions. The accuracy of the test proves that the method is reliable in determining the height of vehicles, achieving 100% accuracy of the detected vehicles. The significance of this work is the design of a vision-based method which can determine the height of the vehicles and is a low-cost alternative to the current costly laser and infrared detection systems.