Effective Video Analysis for Red Light Violation Detection
Maryam Heidari, S. Amirhassan Monadjemi · 2013
Red light Violation detection is one of the exigent branch of knowledge in modern intelligent transportation system, it is preferable to traditional detection methods in operation and maintenance aspect. According to the practice of crossroads and the characteristics of vehicles running red-light, this paper presents the methods of red light runners video detection algorithms sate the real time requirement as well as red-light violation detection with a high accuracy. In addition, they are also capable of tolerating a number of antagonistic but realistic situation such as pseudomotions due to shadows, poor contrast, minimum number of traffic light, pedestrian motions and turning vehicles. Purely from the video, the new methods firstly construct a traffic light sequence, and then detection region to detect moving vehicles and vehicles motions beyond the stop line while the light is red. Our methods are simple and effective, as while as can solve the problems of undetected and wrongly detected vehicles seen in existing algorithms.