Research On Moving Object Detection And Tracking In Complex Condition

Moving target detection and tracking technology is a core research topic in the field of computervision. After studied by domestic and foreign researchers for recent years, a lot of significant resultshave been achieved. However, the complexity of the real environment and the variability of themoving target in practice give rise to the difficulty of moving objects detection and tracking accurately.This paper makes a new attempt based on the existing classical algorithms, which takes real-timeperformance and robustness into account. The main works of this paper are presented as following:(1) Because of such interferences as ghost, and shadow which can not be overcome by ViBealgorithm in moving object detection in practice, a new improved ViBe algorithm is put forward inthis paper. By using inter-frame difference method in preprocessing stage, the true background can begained and the ghost can be removed. And then according to prior knowledge and edge detection inmoving object detection stage, the affection of shadow can be eliminated and the true moving objectscan be got. In addition, with pixel-labeled method, the description of moving object can be achievedand the object can be tracked. Eventually these methods are applied to real-time traffic surveillanceand the experimental results show that these proposed methods have good performance in movingcars detection and tracking.(2) This paper presents an adaptive multi-features particle filter (AMFPF) tracking algorithmwhich combines adaptive sample count particle filter algorithm with some general features such ascolor, texture and edge of targets. The main motivation is to accurately track moving targets usingfewer numbers of particles. This algorithm adaptively controls the number of particles according tothe active contour feature of targets. In the meanwhile, it uses multi-feature fusion method to improvethe accuracy of tracking. The experimental results report that this algorithm not only improves theexecution speed of tracking, but also can track the target accurately and steadily in complex condition.(3) The improved ViBe moving object detection algorithm and AMFPF tracking algorithm areapplied to accurately extract traffic parameters such as traffic flow, vehicle speed and lane occupancyin this paper, thus expanding the application prospect of moving object detection and trackingalgorithm.