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

Abstract

Most traffic regulators uses 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 this project, various image processing techniques were studied and compared. A Matlab GUI was created and programmed to collect statistical data like vehicles classification, vehicles speed, road usage and queue line. Different types of image processing techniques like Background Difference, Edge Detection, Inter-frame difference, Binary image conversion and Quadtree decomposition were used to collect such data. Result from of these techniques will be compared with one another and measured in term of accuracy of the data and timeliness of data collection process.