Dynamic Scene Modeling And Moving Object Detection

As a very important technology of computer vision research, moving object detection can extract interesting moving object from a complex dynamic scenes.Dynamic scenes modeling approach in moving object detection establish the appropriate model to fit the distribution of these changs, by analyzing the distribution of background and foreground modal changes. Scenes modeling approach has generally the assumption that the scenes is simple and static. But the realistic scenes is complex and dynamic. Illumination change, shadow interference, periodic motion and camera shake and other factors will undermine this assumption. Therefore, the study of the dynamic scenes modeling and moving object detection method is of great significance and extremely challenging.From the perspective of statistical probability model and sparse representation,the thesis study dynamic scenes modeling and moving object detection method. by the in-depth study, the main work is as follows:1. Some popular moving object detection methods based on dynamic scenes modeling are deeply studied. By the algorithm analyzing and experimental simulating,the algorithm characteristics are summarized and compared.2. The scale invariant local texture pattern feature and ViBe scene modeling method are deeply studied. A moving object detection method based on scale invariant local texture pattern feature is put forward. The method combines the scale invariant local texture pattern feature and ViBe scene modeling method. To some extent, this method can remove the interference factors such as illumination change, shadow, and improve the robustness of moving object detection.3. The moving object detection based on the sparse representation is deeply studied. In order to reduce the influence of noise, by the elaboration of image sequence and combining the spatial information, a moving object detection method based on spatial-time slice is deeply analyzed. The experimental results show this method can adopt to dynamic changes such as illumination change, rustling leaves, and accurately detect the moving object in indoor and outdoor.