Research On Moving Object Detection And Tracking In Video

Moving object detection and tracking in video sequences is a foundational, core and widely used topic in the field of Computer Vision until now. Moving object detection and tracking technology is related to the field of Computer Vision, Multimedia Technology, Image Processing and Pattern Recognition, Probability and Information Theory, and Artificial Intelligence and so on. So it is an interdisciplinary technique. As a key technology of video data processing, moving object detection and tracking is widely used in many areas from daily life to industry, such as Human-Computer Interaction, video coding, intelligent monitoring, intelligent teansportation, and automated cars. Moving object detection and tracking technology from videos are the premise of the decision-making analysis, simultaneously are the foundation of the intelligent visual system. In recent years, foreign and domestic scholars in the field of object detection and tracking have made a lot of research work and made some research results. However, it is still a challenge issue, and there are some great difficulties such as illumination change, shadow, target part or all of occlusion and deformation, etc. In this dissertation, we study the problems of moving object detection and tracking in video data, analysis several existing models and algorithms, and propose some robust and efficient algorithms for complex environment. Main achievements and contributions are in moving object detection and tracking fields which are as follows:Firstly, we propose a robust method to detect moving foreground and shadow removing in illumination condition. To eliminate the influence of illumination change and shadow associated with the moving objects, we proposed a local intensity ratio model (LIRM) which is robust to illumination change. Based on the analysis of the illumination and shadow model, we discussed the distribution of local intensity ratio. And the moving objects are segmented without shadow using normalized local intensity ratio via Gaussian Mixture Model (GMM). Then erosion is used to get the moving objects contours and erase the scatter shadow patches and noises. After that we get the strength moving objects contours by a new contour strength method, in which foreground ratio and spatial relation are considered. At last, a new fill method is used to fill foreground with holes. Experimental results demonstrated that the proposed approach can get moving objects without cast shadow, and shows excellent performance under various illumination change conditions.Secondly, we propose a moving cast shadow detection method in video surveillance. In this paper, we propose an adaptive cast shadow detection method based on a local intensity ratio vector (LIRV) which is illumination invariant. Then we define the difference of LIRV (DV) and the ratio of LIRV (RV) between shadow and background, and discuss their distribution. The suggested method splits the cast shadows from the foreground by cascading estimators including DV, RV and local spatial relation estimator. Experiments demonstrate that the proposed method outperforms state-of-the-art methods.Tirdly, to solve problems of object occlusion, posture and illumination changes during the process of object tracking, this paper presents a novel several sub objects collaboration tracking method based on adaptive local appearance model. We segment object into many local feature sub-objects, and choose parts of them which are relative significant. In the frame of particle filter, we utilize sparse representation to track every local sub object alone, and estimate object state by tracking results of sub objects. During process of tracking, this paper takes place of unstable local sub objects to ensure the robustness of tracking. Experimental results demonstrate our method outperforms existing tracking methods, and tracking results are more stable and precise especially under the condition of object part occlusion, shift and illumination changes.