Prediction Method for Suspicious Behavior Based on Omni-View Model

摘要

Recendy, CCTV is being applied to prevent crimes. It senses the level of danger as being searching criminal records, a wanted one's montage and so on through mainly Facial recognition. However, it needs additional judging to provide against emergencies, because it cannot predict every criminal situation. In the cause of it, a computer is being fed three-dimensional coordinates from CCTV into a device, and catches not only motion of body or arms but also pattern of hand that were grasped by ConvexHull. And then it predicts suspicious behaviors via judging the movements of an object. Furthermore, to add information about surroundings and location, preventing crimes with more exact judging is the aim on this research.