Research On Intelligent Video Surveillance System Of Multi-Camera

Intelligent video surveillance has been widely used in many fields, such as national defense construction, traffic control, intelligent security, and real-time monitoring of manufacture automation in new high-tech industry; it plays a vital role in the fields of security protection and manufacture. Most of the video surveillance systems still rely on manual operation, which waste resources and influences the efficiency. Concerning about the improvement of real-time and reliable property of the system, an intelligent video surveillance system of multi-camera has been designed and implemented in this thesis. The system integrates the functions of image acquisition, object detection, video recording, data compression, and human tracking.An intelligent surveillance method of hierarchical surveillance has been proposed in this paper. The main camera in monitoring status switches on object detection, data acquisition, and human tracking status, when a sub camera sends a surveillance message to the main camera. The method can reduce large redundance from the surveillance system.The research includes five aspects: image acquisition, object detection, video recording, data compression, and human tracking.(1) Image acquisition: videos and images are captured to computer through cameras, then the capture parameters of every camera will be set up and the window will be displayed dynamically.(2) Object detection: the research on object detection is mainly based on the method of interframe difference, background subtraction and optical flow. The motional object detection is implemented by the interframe difference method.(3) Video recording: the research on video recording is based on the technology of DirectShow and VFW, and video recording is implemented by the two methods mentioned.(4) Data compression: the research on image compression method is based on JPEG encoder, and video compression method is based on AVI-ICM. Then data compression of the system is implemented by the two methods mentioned.(5) Human tracking: considering the characteristics of the system, theory of Mean-Shift is researched in emphasis, Extended Kalman Filter is added to the theory to estimate and forecast the status of human motion, and the people tracking based on improved Mean-Shift algorithm is proposed.A software of intelligent video surveillance of multi-camera has been designed, which includes five parts mentioned above and integrates various algorithms to surveil the monitoring scene in real-time by controlling the cameras automatically.