Automatic Detection and Tracking of Ship Based on Mean Shift in Corrected Video Sequences

Robust real-time ship detection and tracking for visual images automatically has become one of the crucial requirements for numerous situations. In order to improve the performance of automatic ship target detection and tracking system, a novel method based on mean shift is proposed for automatic detection and tracking of ship in the corrected video sequences. Firstly, video frames are corrected based on real-time attitude of the imaging equipment in order to decrease the drift of the ship target from frame to frame. Secondly, sea horizon is extracted based on Ostu algorithm and Hough transform. Thirdly, the ship location is detected based on the detection of grayscale peak. Finally, the ship is automatically tracked by using the mean shift algorithm. The result shows that this method has not only improved the robustness of the system against the shaking while tracking the ships, but also achieved a high success rate of tracking ships automatically in corrected video sequences.