Paper #4

A Novel Traffic System for Detecting Lane-Based Rule Violation

Md. Azmol Fuad, Faed Ahmed Arnob, Abu Tahir Nizam and Md. Motaharul Islam

Abstract: In recent years, there have been a rise in the number of problems in the existing traffic management system particularly in the developing countries such as Bangladesh, India and Vietnam. Due to this, many accidents have occurred every now and then. Violating the traffic rules such as unpermitted change of lanes and over speeding are the two main reasons for increased number of accidents. In this paper, an attempt is made to solve the problem using Raspberry-pi and OpenCV contour detection technology. A prototype device has been developed to solve the problems regarding lane-based rule violations. The device will be installed in traffic surveillance cameras which will be connected to the metropolitan traffic servers. The device will communicate with the server via Gigabit Ethernet. It will also increase the time efficiency and reduce the manual monitoring cost. Moreover, it will help the traffic management department to find the person responsible for traffic rule violation and assist them to apply the laws strictly. The main contribution of this article is to develop a device that will detect any kind of unpermitted lane changes by any vehicles and identify the license plates. The proposed model has about 80% accuracy according to our evaluation.

Keywords: OpenCV contour; License Plate Recognition; Hough Line Transform; KNN Algorithm; Canny Edge detection; Gaussian Blur; Network Attached Storage.