A Novel Approach for Night-Time Vehicle Detection in Real-Time Scenario

Intelligent transportation systems and computer vision have intensively been studied in the past decades since the need for intelligent transportation system increases as the growth of vehicles is increasing rapidly than ever before. Traffic surveillance is the important feature in intelligent transportation system which involves in detection and counting of vehicles. In this paper, a night-time vehicle detection system is proposed for effective detection and counting of the vehicles during night-time with the help of the headlights and also classification of the vehicles is carried out in this system. This system involves in processing the night-time traffic video in MATLAB, where image segmentation, background subtraction, blob analysis are done to detect and pair the vehicle headlights for the counting of vehicles. Different templates of headlights are created, and comparison is carried out with the headlight of the vehicles present in the traffic video to classify the types of vehicles present. The above algorithm for night-time detection and counting of vehicles is also implemented in system on chipboards, so that it can be effectively used for intelligent transportation system purposes.