作者: James J. Clark , Vinod Nair
DOI:
关键词:
摘要: This paper describes an automated visual surveillance system that detects suspicious human activity in a scene. The is designed to: 1) detect and track people the scene, 2) recognize “normal” activities 3) anomalous by finding sufficiently large deviations from normal patterns. stochastic time-sequence recognition framework of Hidden Markov Model (HMM) forms basis anomaly detection. We have implemented to monitor office corridor real-time using Pentium III machine running Windows 2000. results show correctly classifies examples identifies mock break-in attempt as activity.