Research On Detection And Recognition Of Abnormal Behavior Based On Indoor Scene

Intelligent monitoring system has attracted wide attention because of its advantages of all-weather, uninterrupted, low false alarm real-time monitoring. And the key technologies of the target detection, target tracking and behavior recognition are the focus in the scholars' research.In view of the indoor fixed scenes, the technology of target detection, target tracking and behavior recognition were studied thoroughly, and the existing problems of them had been solved. Target detection part of this paper combined background difference method based on Vi Be and the frame difference method, selects the better target detection result by setting threshold of illumination change, which could solve the problem that Vi Be algorithm cannot detect moving target accurately when light changes. The target tracking part used motion estimation of Kalman filter to improve Camshift algorithm, the Bhattacharyya distance and occlusion rate was used to judge whether target was occluded or not and obscured degree. It could effectively solve the unstable tracking problem when target was occluded. Abnormal behavior recognition part propose d a kind of improvement human abnormal behavior identification algorithm based on template matching, improved Hu invariant moments and image motion characteristics was combined to form feature vector, the Hausdorff distance was used to calculate the similarity between test behavior and template behavior, and the corresponding threshold to determine whether the test behavior was abnormal behavior or not.Experimental results show that the improved target detection, tracking and abnormal behavior recognition algorithms were effective, and the recognition rate of abnormal behavior was improved.