One Research Of Detection&Identification And Simple Motion Analysis For Dynamic Body Contour Based On The AVI Video

Accompanied by the emerging computer vision technology and in-depth study, its applications such as intelligent control system, virtual reality systems, and digital video coding technology tend to flourish. Full-respected in the information and technology innovation for nowadays, special and very important as a national law enforcement agencies, reliability and intelligence of the prison monitoring system is particularly critical. Prison monitoring system to monitor the main are prisoners, and monitoring the scene includes prisoners’living room, visitation rooms, canteens, important channels, access control and etc. Complex and special monitoring of the subject and the monitor scene, raises extremely demanding requirements on the function of prison monitoring system. Based on this, starting by the theory and practice, for simulating the prison environment,this paper conducted superficial exploration and research on the video frame pre-processing, edge detection, extracting and tracking of moving objects, identification and counting the number of human targets and other key technologies etc. The main contents are as follows:1.Since the video frame pretreatment technologies are widely ranged relatively, for actual situation of the prison monitoring environment, the article focuses on the following processing:Through gray-scale processing of color images, gray image binarization, we aimed to reduce some unnecessary pixels, thereby reducing the engineering workloads; By histogram equalization, image denoising, image sharpening, we completed smoothing and filtering of the image frame to prepare for the next edge detection.2.For the image frame edge detection,we used a variety of edge detection algorithms to compare the experimental simulation results. The edge detection algorithms used include classical edge detection operators (gradient operator, Roberts operator, Sobel operator, Prewitt operator, Laplace operator, and Wallis operator) and modern edge detection operators (edge detection algorithm based on BP network, Pal.King fuzzy edge detection algorithm and the advanced Pal.King detection algorithm by the author). 3. This paper used the neighborhood frame difference method to complete the moving object extraction, and used the double Kalman filter for tracking of the moving objects.The simulation results showed that the neighborhood frame difference method could effectively extract the moving object silhouettes, and the double-adaptive Kalman filter, compared to the classic Kalman filter, was better able to accomplish the target tracking with adaptive-fast convergence for the situations such as target mutations.4. The article studied on the dynamic human target identification, the number of statistics for human targets and simple motion analysis. On the basis of the research discourses by ancestors, the paper proposed an integrated algorithm (the human recognition based on the body contour aspect ratio, and the head characteristics, the section type statistics to calculate the body number), through the experimental simulations, and data analysis, achieved the purpose of the research in the mock environments of the prison monitoring system.