A sensorless red light violation detection system using image processing

Abstract

A new sensorless red light violation detection system is developed to detect vehicles violating the traffic red light using image processing techniques. The system is capable of taking the picture of the vehicle violating the red light rule using only a single camera. The basic concept on which the vehicle detection using image processing is based is known as frame differencing where a comparison is made between the current picture frame and the previous one. Once the vehicle is detected, it can be tracked through the video frames for taking the picture of its license plate. There are two main problems that can arise while detecting vehicles using image processing that can result in false triggering of the camera. One is when a non-vehicle object such as a bird or a pedestrian crosses the border line while the vehicle is waiting at the red light and the other problem arises due to the changes in daylight. False detection can be avoided by using the interframe difference technique.