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

• Main Author: Ankit Jaggi
• Other Authors: Mohammed Yakoob Siyal
• Format: Final Year Project
• Language: English
• Published: 2009
• Subjects: DRNTU::Engineering::Electrical and electronic engineering::Electronic systems::Signal processing
• Online Access: http://hdl.handle.net/10356/18383
• Institution: Nanyang Technological University

Nowadays, there is a huge demand for real time traffic analysis due to the enormous number of vehicles present on roads. This has led to the need for devising control systems which can allow us to collect the statistics on the amount of traffic flow. The author has researched and devised various algorithms to get relevant data for road traffic analysis, in order to obtain valuable information which can be later used for improving traffic management and its efficiency. The first step was obviously to obtain usable images from the real time traffic videos as recorded by camera equipment and then process them using various image processing techniques. In this project, the proposed method is to use window-based method with various segmentation methods such as background difference, interframe difference, edge detection, binary image conversion and quadtree decomposition. The author has also studied other techniques like HSI, thresholds and use of GPS systems to improve the algorithms developed. The main purpose was to investigate, develop and implement various algorithms to study the ‘quality’ factors of road traffic like class of vehicle, lane speed and road usage implementations. The project also required for the author to study some quantity factors like vehicle count in order to gain a better understanding of traffic analysis concepts. The algorithms developed can be tested on sequence of images under various conditions and thereafter, study the changes due to external factors like weather, time of day and the lighting of the area. This helps in identifying the image processing techniques, which are most suited for studying traffic management and their constraints. The results obtained are pretty encouraging but there is still room for improvements such as removing shadows on vehicles, vehicle overlap issues and using ‘smart’ new technology like GPS and neural networks.