Date of Award
Master of Science (MS)
Ogle , Jennifer
Choudhury , Ronnie
This thesis presents an evaluation of the accuracy of a novel computer vision traffic sensor - developed by the Clemson University Electrical and Civil Engineering Departments - capable of collecting a variety of traffic parameters. More specific, the thesis examines how the camera height and distance from the travel way affects the accuracy. The details of the quantitative and qualitative evaluations used to validate the system are provided. The parameters chosen to evaluate were volume, vehicle classification, and speed. Experimental results of cameras mounted at heights of 20 and 30 feet and a lateral distance of 10 and 20 feet show accuracy as high as 98 percent for volume and 99 percent for vehicle classification. Results also showed discrepancies in speeds as low as 0.031 miles per hour. Some issues which affected the accuracy were shadows, occlusions, and double counting caused by coding detection errors.
Graham, Robert, "EVALUATION OF A COMPUTER VISION TRAFFIC SURVEILLANCE SYSTEM" (2007). All Theses. 290.