Application Research Of Intelligent Video Analysis Technology In The Financial Network

With the development of social economy and technology,security problems such as banks and other financial systems are facinghuge test, and how to ensure the safety of these financial place is moreand more concerned and draws attention from all walks of life. Theemergence of the intelligent monitoring system has played a significantrole in this issue, and now it has become an important research directionin the field of computer vision. Due to the intelligent video monitoringsystem with the help of the computer vision, image processing andpattern recognition technology, it is well-behaved to analyze andrecognize the abnormal behavior in a monitor scene, and the target of theabnormal behavior can be fed back to relevant departments in real-time,then the departments may pay more attention and handle it. Therefore, thesecurity forces need not keep eyes at the monitor display like it wasbefore, and this will greatly save manpower, material resources andfinancial resources. In the meantime, the real-time performance andreliability of the monitoring system can be both improved.In this paper, the method of intelligent video monitoring system toidentify abnormal behavior such as abandoned objects, wandering,masked in special scenarios is proposed. Moving objects detection is thebasis and key to the abnormal behavior identification. This paperintroduces some common methods which concern moving objectdetection, then analyzes the advantages and disadvantages of variousalgorithms, and finally chooses the background difference method formoving object detection. In view of the background difference method,several kinds of background updating method are included, the Gaussianmixture model can solve the impact of the background changing (lightgradual change, light intensity, etc.) to get the background model, soGMM is selected to update background; After the detection, usingmorphology to deal with noise in the process of target detection and get afull set clear goal.Under the condition of static single camera, the combination ofbackground model which is constructed by GMM and the background difference method is used to detect abandoned objects, and the algorithmis proved to be quick and efficient. When testing for wandering behavior,the first should get the external contour of the object, and then byobtaining the center of the moving object external contour track tocompute the distance, displacement and direction of the moving object,finally concludes weather the moving target is wandering behavior or not.For moving object with a masked face, the moving object region is set tothe interested region firstly, and then YCbCr color space ellipse skincolor model is used to detect the skin color in the specific scope of theinterested area, which can greatly shrink the search scope and reduce theamount of calculation. Finally analyzing the connected component of theskin color detected, through calculating the area of the connected regionwhether facial block phenomenon exists.The paper has done a lot of related experiments and other methodsare compared. The experimental results show that the method’scalculation is simple, and greatly reduce the complexity in time and space.At the same time it shows high-efficiency real-time and recognition rate,and is suitable for real-time intelligent monitoring system.