Abstract: Mathematical model of each component of target tracking system based on dual-pan-tilt camera was established. The rotation angles both in horizontal and vertical direction of each pan-tilt and the relative position of two cameras were adopted to estimate the real-time position of moving target with Extended Kalman Filter (EKF) tracking algorithm. A new real-time calibration method for pan-tilt camera was proposed to deal with the calibration problem that the external parameters of camera changes as the pan-tilt rotates in the process of camera modeling. In addition, the target’s image feature, extracted from the projected point on the camera image plane of estimated position of the target, was used as feedback information of visual servo system to control the pan-tilt to track the trajectory of the moving target. The Matlab/Simulink simulation system was built and the results show the scheme feasibility and effectiveness. It also provides a test platform for further research in the controller design of visual servo system.

Key words: pan-tilt camera, binocular stereo vision, target tracking, EKF, visual servo