Research On Algorithm Of Moving Object Detection And Tracking Methods In Intelligent Visual Monitoring

With the continuous development of information technology and increasement of social need, the intelligent video monitoring has been widely used. The foundation and key in intelligent video monitoring is detecting and tracking of moving targets, it has integrated image processing with pattern recognition and artificial intelligence and some other research. The detection and tracking of moving targets has important applications in national defense and military, process controlling, medical image processing and social security and other fields.Detection and tracking of moving targets is one of the hot spots and focus in current research. It has been reached a lot of achievement during the past years,however, there also been many problems and shortcomings. This thesis will analyze and improve this method based on the previous research. The detailed research work is as follows:On detecting moving targets, this thesis does a careful research and analyzes some fundamentals about optical flow method, background difference method and frame difference method. It has made a comparison of these three methods through simulation experiment, and analyzed the result. On that basis the thesis pays more attention on the background difference method research, and does a further research on background model building. Besides, the paper puts forward the adaptive update rate Gauss mixture model on improving the classic Gauss mixture model. And then, it proposes a new detect method that Three-frame Difference Method combined with improved Gauss mixture model, and does a simulation experiment for the new detect method. Comparison between new and old method, it comes out that the new method has been more effective in detection.It firstly makes an analysis in Mean Shift, Camshift and Particle Filter when doing a research in moving targets tracking. This study put forward fusing multi feature information in Mean Shift and Particle Filter according to the advantages and shortcomings of Mean Shift and Particle Filter. Secondly, using the new shift in targets tracking and having a experiment on it and analyzing the result. Compared with the classic Mean Shift and Particle Filter experiment, the multi-featured Mean Shift and Particle Filter has a greater precision and effectively improved the tracking result.