Researches On Key Technologies Of The Multi-Camera Collaborative Tracking System

With the popularity of the computers and the improve of the capacity of the computer storage, network transmission and data processing, people are becoming more and more dependent on computers as the computers are used more and more widely in our daily life. At the same time, with the developing of the video storage and processing technology, people rely on the computers to do the 7 X 24-hours real-time monitoring, detection and tracking works or some tough tasks in the complex environment which people can not take participate in directly increasingly.Although the intelligent video tracking technologies make people get rid of the heavy monitoring and identification works nowadays, they all have a common character that each camera is a relatively independent system which requires people to intervene in the synergy between monitoring, identification and tracking. Most current video tracking technologies only do the simple recording of the interesting tracking areas to backup for the indexing.This paper is mainly introducing the researches on the key technologies of the multi-camera collaborative tracking system. There are three parts constitute the multi-camera collaborative tracking system. First, Main System which contained a camera and a master program which responsible for monitoring the scene, detecting the moving objects and communicating with Server and Subsystems. Second, Subsystems: could have more than one which contained a PTZ controlled camera and a sub control program per subsystem, responsible for receiving commands from the Main System, tracking and display the moving objects in the scene in real-time and sending the result to the Server at the same time. Third, Server and IE Client:Server is responsible for transmitting the relevant cameras'information received from the Main System and the Subsystems to the IE Client and transmitting the users' commands accepted from the IE Client inputted by the clients to the relevant Subsystems or Main System simultaneously in order to achieve the specific real-time Subsystems or Main System control and remote interaction.In the process of the video object tracking, color information analysis is the main method in monitoring the scene and tracking the moving objects. Frame difference, smooth filtering, CAMSHIFT and particle filter are the common algorithms involved in the video object tracking. And use the TCP/IP protocol to accomplish the communications between the cameras to cooperate to get the work done. In order to improve the efficiency of detection and tracking, we use the different algorithms in the scene detection and the moving objects tracking.We test the key algorithms in multi-camera collaborative tracking system. There are three computers constitutes the simulation experimental environment:one computer with the Main System and a camera; the other computer with the Subsystem and a PTZ controlled camera; and the last computer with the Server and IE Client. Monitoring scene is simulated by the moving objects in the projector's projection.