An Intelligent Surveillance System for Crowed Abnormal Detection

摘要

In this paper, an intelligent surveillance system for emergency detection is designed for campus. The framework of the system includes following steps. First, the foreground extraction is used with background difference and three-frame-difference methods to get the moving objects. Second, based on canny operator, an edge detection is employed to eliminate the effect of brightness. Third, to enhance the accuracy of the results at non-vertical visual angle, the image is divided into limited layers to detect separately after edge detection, based on pixel features. In this way, the crowd density information can be obtained. Experimental results show the effectiveness of the designed system.