Automated Traffic Monitoring Using Image Vision

The process of traffic monitoring is predominantly carried out manually in our country. Traffic monitoring encompasses a set of stringent rules to be followed while ensuring that there are no traffic jams at the juncture of roads. In this paper, a novel method is proposed to automate the traffic signal lights with the assistance of multiple CCTV cameras connected over the Internet to survey various roads at the junction. The process comprises two primary phases: Vehicle Detection System and Traffic Scheduling Algorithm. Vehicle Detection shall be carried out in Digital Image Processing (DIP) by applying a simple kernel-based Edge Detection in Spatial Domain followed by an algorithm to detect the perimeter of closed figures while simultaneously applying the concepts of Machine Learning to classify the vehicle type into the following categories of motorcycle, light motor vehicle, and heavy motor vehicle. Subsequent processes are carried out in a novel Traffic Scheduling Algorithm through the help of a hybrid Round Robin having a dynamic time slice obtained by using Longest Remaining Job First to periodically update the traffic signal lights to relax the traffic. Instead of turning on the green light for a fixed amount of time, the duration will be managed dynamically based on the amount of traffic in each road. Thus, the proposed system is aimed at reducing the jamming of roads by a huge extent.