Design and development of image processing algorithms for traffic applications

Abstract

Most traffic regulators use surveillance cameras to monitor road traffic patterns, perform post-accident investigation and gather traffic statistics. For statistical data collection, it heavily involves manual counting of vehicles through the surveillance footage which is tedious and unreliable process. With the aid of autonomous traffic analysis software, it improves the process of road traffic data analysis, and it provides more accurate and timely road traffic information.

In recent times, there has been an increase in the demand for the real time traffic analysis of vehicles moving on the road. This in turn has led to an increasing need for the development of control systems which can gather statistical data related to the flow of traffic. In this project, various algorithms for traffic data analysis were researched and implemented, so that it can be applied for the measurement of real time traffic.

In this project, the proposed method is to use Window-based method with five segmentation methods such as background difference, edge detection, inter-frame difference, Quadtree decomposition, and binary image conversion.

A Matlab GUI is developed for the simulation. The traffic data algorithms are designed for vehicle count, vehicle speed and road usage. Window based detection is adopted to detect pixels within the detection zone that will trigger the traffic data algorithms.

Videos of real time traffic at different times were first captured with a digital camera. These videos were then converted into image frames which are later processed using different image processing techniques.