Automated Counting and Tracking of Vehicles

摘要: A robust traffic surveillance system is crucial in improving the control and management of systems. Vehicle flow processing primarily involves counting tracking vehicles; however, due to complex situations such as brightness changes vehicle partial occlusions, traditional image segmentation methods are unable segment count vehicles correctly. This paper presents a novel framework for vision-based tracking, which consists four main procedures: foreground detection, feature extraction, analysis, counting/tracking. Foreground detection intends generate regions interest an image, used produce significant points. Vehicles achieved by analyzing clusters As testing on recorded videos, proposed verified be able separate occluded number accurately efficiently. By comparing with other methods, we observe that achieves highest occlusion rate accuracy.