Vision-based system for road traffic applications

• Main Author: Luo, Yuefen
• Other Authors: Mohammed Yakoob Siyal
• Format: Final Year Project
• Language: English
• Published: 2017
• Subjects: DRNTU::Engineering::Electrical and electronic engineering
• Online Access: http://hdl.handle.net/10356/70837
• Institution: Nanyang Technological University

It is a conventional topic to apply image processing algorithms to a sophisticated road traffic vision. However, there are various vulnerabilities to be resolved. This project practices the theoretical image processing algorithms in an object-oriented development environment. Using of MATLAB is a strategic approach to simulate vision-based system locally before it can be planted to a real-world traffic situation. After a prior investigation, the program development proceeds with structure modeling, source coding, GUI rendering, data binding and development-side testing. As a result, the project demonstrates several stand-alone applications after compilation from MATLAB. Its integral parts include: consolidating the primary image processing algorithms into a generic image parser; recognizing vehicle registration plate thereby segmenting characters using OCR; recognizing road traffic sign by merging HSV colormap and feature matching techniques; monitoring the traffic congestion by detecting and counting vehicles on each lane. The project approximates the reliability of adopting vision-based processing algorithms in road traffic systems through interfacing with its video surveillance. Hence, the project has been carried out successfully. This thesis collects experimental findings from the project, and illustrates the methodologies of image processing that can be implemented in road traffic applications.