Vision-based system for qualitative road traffic data analysis

Abstract

In today’s world, with the population and living standards increasing the minute, the demand for vehicles is increasing. There is an influx of vehicles on the road resulting in a rise of traffic. The most common way to monitor road traffic is through surveillance cameras. But counting the vehicles manually is a very cumbersome and time-consuming process. Therefore, there is a growing need to develop an efficient system to monitor the real-world traffic situation and analyse it according to the requirements. The main aim of this project is to investigate, design and implement image-processing algorithms in order to meet the growing requirements for traffic analysis such as, monitoring the flow of traffic to collect relevant data. For the purpose of this project many imageprocessing techniques have been implemented such as, Binary Image conversion, Background Difference, Inter-Frame difference, Edge Detection, Optical Flow and Quadtree Decomposition. These techniques are implemented to provide a quantitative analysis of the road traffic situation. These techniques have been used to implement several algorithms such as calculating the number of vehicles on the road, the vehicle speed, classifying the vehicles into different categories depending upon their size and calculating the percentage of the road used. All these algorithms have been implemented using MATLAB programming by integrating them into a MATLAB Graphical User Interface. This GUI takes into consideration the various requirements of the user, for example selecting the number of frames to be evaluated by the program, selecting the technique to evaluate them and more. Results obtained by these various techniques have been obtained and compared based on different criteria. The author has also performed research on other methods available for quantitative road traffic analysis and has also managed to provide certain recommendations for further improvement on this project.