Design and development of image processing algorithms for quantitative road traffic data analysis

Abstract

Vision based road traffic analysis plays a vital role in modern traffic system. It introduced automation for image processing that is capable of collecting statistical traffic data while obsoleting manual counting. This allows agency to monitor traffic patterns and implement relevant measures to reduce congestion, especially during peak periods. In this project, various edge detection algorithms are used to study the surrounding light sensitivity, focusing on the transition from day to night for road traffic data analysis. 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. Other image segmentation methods such as background difference, inter-frame difference, binary image conversion and quadtree decomposition are incorporated for comparison in the initial analysis.