Optical Sensor Network On Motion Detection And Object Tracking System

Motion detection and object tracking is an important research field of machine vision. In most tracking systems based on background subtraction method, the optical sensor need to be fixed, it may definitely limit the flexibility of the system. In order to eliminate the dependency on the background model and increase the anti-interference ability, a tracking system with movable optical sensors network is proposed. The following is the short description of the system.1) ROI selection:Since the dynamic tracking system is sensitive to all the moving objects, the user need first select the interested target. After that, the velocity, direction and acceleration information can be acquired.2) Object detection:After subtracting two adjacent images and threshold the result image, the white part in the result image will be marked as motion of the target.3) Object tracking:After retrieving the distance from the motion to the central point of the image, can the driving angles of the two servos be calculated. According to the angle information, the two servos can be driven to set the camera's line of sight cross the target so that the target can not move out of the camera's field of view.4) Prediction of target trajectory:If the target is temply blocked and the camera can not see it, the image subtraction processing method will fail to work. In this case, a predictor can be used to evaluate the next step of the moving target. Besides, because of the other disturbances, the system is possible to track the wrong object. If this happens, the predictor can also examine the tracking result to avoid a wrong tracking.5) Data transmission:At least two cameras can form the 3D information of an object. Furthermore, multi cameras can form a decision making mechanism. Through data transmission, the data from all the cameras can be collected together to achieve a high trustable data to realize the high efficiency of object tracking. The system realized a prototype of the motion detection and object tracking system based on optical sensor network. It successfully achieved the ability of object self-tracking and can filter most of the disturbances. It is a new trial of tracking method.