Research, Based On Image Analysis Of Self-service Banking Scene Of Intelligent Security Technology

Automatic teller machine has brought to the life with a lot of convenience, but more and more law breakers gaze on these public facilities and take a variety of criminal falsifications to violate people's property. In this dissertation, a research has been made on camera tampering detection, abandoned objects detection and illegal notices detection for the modus operandi of criminal falsifications to automatic teller machine.First, the image characteristic function based on Harris corner and the algorithm of camera tampering detection has been built based on the features of the scene of self-service bank. When the scene changes, the values of the adjacent frames'characteristic function will also change, one large amplitude crest that show the difference value of the adjacent frames' functions will appear, so the behavior of camera tampering can be detected, and the feasibility and effectiveness can be proved through the simulation experiments.Second, an improved algorithm for changed region detection in the monitoring scene based on the method of background subtraction has been built in view of the large pedestrian volume in self-service bank, this algorithm can avoid the situation that identify the people in the bank as the target. The contour, the area and the rectangular similarity of the candidate region have been extracted through the technique of chain code tracking, interference region can be filtrated and abandoned objects can be detected effectively through restriction of the shape parameters, and the experiment result shows that the method is feasible.Third, four features based on statistics of edge image, corners of image and connected regions of image are built through the analysis of low resolution document image, and the distributions of the features are analyzed. The samples are tested by classifier based on SVM (Support Vector Machine) and threshold, the threshold and the parameters'value are ascertained by GSM (Grid Search Method). The results of the experiments prove that the two methods can both detect the illegal notices in self-service bank efficiently. Finally, the two methods are compared and the result shows that according to the features got in this paper, the method based on threshold is better than the method based on SVM from the view point of engineering application.