UKF-EC: Combining the Unscented Kalman Filter and the Maximum Weight Algorithms for Moving Target Tracking in Marine Radar Images

Abstract

For marine radar target tracking, the real data differs from the simulation data, the tracking accuracy is poor and the track is easily changed due to interference from sea clutter. UKF-EC algorithm is proposed: Combining the unscented Kalman Filter and the maximum weight algorithms for Moving target tracking in a marine radar image. The algorithm first completes target detection via the target contour feature, then improves tracking accuracy via the designed UKF state estimation module, the data association module completes the matching between the measuring boxes and the prior boxes, and the track discrimination module decides whether to terminate or restart the track based on the history and future time. Finally, it increases the marine radar target’s tracking accuracy and makes it possible to keep track of the target even when it is not detected, vanishes and reappears, interferes with sea clutter, etc. Meanwhile, the track ID won’t change at random in the interim. The experimental findings demonstrate that, in comparison to other approaches, the UKF-EC algorithm put forth in this paper can successfully complete target tracking on a real dataset, the target state estimation is more useful, and it is difficult to change the ID of the track when the detection frame is off. Additionally, the system presented in this paper processes each frame of an image from detection to tracking in just 89.6 ms, enabling accurate and real-time monitoring of moving targets on the sea surface.