Research On Vehicle Detection And Tracking Based On Video

With the development and progress of society,Intelligent Transportation System(ITS)attracts wide attention worldwide.Intelligent Transportation System makes a huge difference on road monitoring and freeway management.Research on vehicle detection and tracking based on video has always been hotspots and difficulties in ITS.This paper mainly focuses on the following points.(1)We first research moving target detection based on background modeling.This paper analyses Vibe algorithm's advantages and disadvantages and proposes improvement measures to suppress the ghost.Meanwhile,to reduce the influence of cast shadow effectively,we propose a fusion algorithm which combines the feature of LBP with non-parametric HSV histogram.(2)This paper also studies multi target tracking algorithm and introduces the fundamentals of several common methods such as kalman filter,optical flow and particle filter.We propose an enhanced histogram of the oriented gradient combined with HSV color space based on particle filter.It can work well under normal light condition and the low contrast environment.(3)Due to the phenomenon the interaction and overlapping between vehicles,the appearance of vehicle will be different in two adjacent frames Occlusion makes it difficult to associate the same vehicle extracted from the frames.We employ the occlusion detection and occlusion segmentation to refine our system,which are based on one-dimensional maximum entropy and the least square ellipse fitting.This paper carried on simulation of vehicle detection and tracking in different situations.Statistical results show that the robustness and efficiency of our framework which can handle shadow and occlusion effectively.