The Research On Detection Of Abandoned Objects In Video Sequences

With the Enriching of video contents and the deep of video applications, based on applications of video sequences has gradually become the mainstream on multimedia. So, the Detection and segmentation based on video has become one of the hotspots about the worldwide research and product development. Based on deep summarize and analysis of existing detection method about moving targets, the paper focuses on the detection of abandoned objects when video camera is fixed. Paper's work can be divided into five parts: Firstly, article introduce the purpose, meaning and application of the research based on video sequences, and analyzed the abandoned objects detection's current situation and future trends. Secondly, in then second section, there simply introduce basic principles about video analysis. Then study in the video analysis technologies applied in target detection, include tasks and processes of moving target detection and commonly used methods of image segmentation. Depth study of two commonly used methods of target detection: Frame difference and Background difference. Separately described the basic principles, advantages and disadvantages. Furthermore, in third chapter, there discuss in detail about the Principles, Purpose, steps and algorithm constraints of abandoned objects detection's algorithm. For the feature of abandoned detection, proposing a abandoned detection algorithm which based on scene classification. After detected target objects in different scene, Useing center-judged method to judge the abandoned objects'feature, and then mark up these objects separately. In order to lower the false alarm rate and improve the accuracy of abandoned objects detection. Then, in forth chapter, we choose a video sequences which contain three scenarios to test the feasibility and practical of abandoned objects detection. Then analysis the result of experimental and point out the advantage and weaknesses. Finally, summary the full article and explore the improvement direction of abandoned objects detection in video sequences.