Long-term Adaptive Tracking for HFSWR Vessels Combined with ELM

Due to highly maneuverable vessels, dense channels and strong clutter, intermittent track segments are common in large-scale marine monitoring. To solve this problem, we propose a long-term continuous tracking method for high frequency surface wave radar (HFSWR) based on an adaptive filtering algorithm combined with an extreme learning machine (ELM). Via analyzing the features of the vessels and tracks and lots of simulations, we select average velocity, average acceleration, average heading angle, average curvature, ratio of the arc length to the chord length, and wavelet coefficient to compose the multidimensional feature vector to train and test the ELM. Field experiment results show that the proposed method has better performance than conventional algorithms with the correct association probability of 94.7%.