Research On Intelligent Video Surveillance And Retrieval System

With the widely use of surveillance cameras, people’s interest for intelligent video surveillance and retrieval are more and more strong. In this thesis, we give a detailed analyzing on the constituent of intelligent video surveillance and retrieval system and provide some methods to key problems in this framework. The main research works and contributions of this dissertation can be summarized as the following:1. We present a multi-feature background subtraction method which follows the ViBe paradigm and using both color feature and SILTP operator to model the background. The method based on HSV color space is used as an assistant measurement for shadow detection. The experimental results show that our method can combine the hypothesis of texture and color to get better subtraction and shadow detection results.2. An object tracking method based on Compressive Tracking and Kalman filter by using the result of background subtraction is presented. The combination of these two methods improve the robustness and accuracy of object tracking.3. We analysis and point out the structural model of surveillance video, proposing video surveillance system based on objects and events. We extract the HSV visual color to suppose the visual color retrieval. We extract the MPEG-7domain color feature and edge histogram feature to suppose hierarchical and fast image retrieval.4. We analysis the possibility and necessity of combining video surveillance system and video retrieval system, combining them to fit application requirement and reduce redundant. We present an intelligent video surveillance and retrieval system based on web technology which can do real-time monitoring and video retrieval.