Real-Time Follow-Up Tracking Fast Moving Object with an Active Camera

Abstract

A method for real-time follow-up tracking fast moving object with an active camera is proposed. The mean shift algorithm shows an excellent performance on robust and fast object tracking; however, it is prone to fail when tracking a very fast moving object. This paper addresses to solve the problem. At first mean shift based on the color feature of the object is introduced. The object maybe can't be tracked when the motion of the object or the motion of the active camera is very fast. In order to overcome this difficulty, Kalman filter is introduced to predict the location of the initial search window of mean shift. In order to reduce the disturbance from the background and to reduce the computational consumption, a region of interest (ROI) is introduced. To follow-up track the moving object smoothly, a closed loop control model based on speed regulation is applied to drive a PTZ camera to center the target. The results of our experiments show that the active camera can follow-up track a fast moving object smoothly. The system is computationally efficient and can run in real-time speed.