Automated road traffic census using image processing technique

• Main Author: Ng,, Hooi Sin
• Format: Final Year Project Report
• Language: English; English
• Published: Universiti Malaysia Sarawak, (UNIMAS) 2013
• Subjects: QA76 Computer software
• Online Access: http://ir.unimas.my/id/eprint/39277/1/NG%20HOOI%20SIN%20%2824%20pgs%29.pdf; http://ir.unimas.my/id/eprint/39277/4/NG%20HOOI%20SIN%20%28fulltext%29.pdf; http://ir.unimas.my/id/eprint/39277/

This thesis proposes the development of an automated road traffic census prototype that is applicable to replace the current manual counts approach by applying image processing techniques. The developed system can be divided into three successive phases: the first phase is video frame preprocessing where a filtering technique using mask image is applied by defining the region for analysis in order to reduce computational complexity. The second phase is vehicle detection stage where the foreground moving vehicle is detected and identified using a combination of value and saturation (CVS) background subtraction technique. The third phase is vehicle counting and classification stage. In this phase, information of bounding box such as center point, width and height of the identified vehicles are used in implementing the counting and classification algorithm. Vehicles are counted when they passed through the defined lines and classified based on their dimension. The parameters used in the algorithms are discussed in detail in the report. The performance of the system is tested on four pre-recorded videos under different environment. The experiments' results show that the system has achieved an overall accuracy of 95% in detecting vehicles and accuracy of 82% in counting and classifying vehicles with a low amount of processing time.