Vehicle information is invaluable in many transportation issues. Vehicle detection and feature extraction is the process of inspecting vehicles and can be used to classify vehicles. Current systems for automatically classifying vehicles have deficiencies and need to be improved.; This thesis introduces a novel, model-based vehicle classification system using computer vision, pattern recognition and image processing (CVPRIP) technologies. In this system, two-dimensional (2D) models are composed with the length, width, and height of the vehicles as key features. The captured images are preprocessed and segmented for vehicle classification. The system was tested using various images captured by the highway traffic control office of the Utah Department of Transportation (UDOT). Because the images were captured with random orientation, they were worse than the data set used by other algorithms. The experiments' results show that the performance of the system is better than those of the existing video-based vehicle classification systems.
展开▼
