An Adaptive Ionosphere Clutter Suppression and Target Detection Method for HFSWR Maritime Surveillance

High-frequency surface wave radar (HFSWR) as a maritime surveillance facility faces the tough challenges. One of them is that various unwanted echo components constantly destroy the detection performance of HFSWR, especially the dynamically changing ionosphere clutter. This article proposes an adaptive ionosphere clutter suppression and target detection algorithm (AICSTD) based on improved Higher-Order Singular Value Decomposition (HOSVD) and deep learning network. First, we analyze the multiple-dimension characteristics of different types of ionosphere clutter echo data. Then, refer to the multi-channel HFSWR echo data structure, we apply simplified HOSVD and orthogonal subspace projection to eliminate the unwanted echo component, meanwhile, preserve the target echo. The key parameter of simplified HOSVD is determined by the mean structural similarity (MSSIM) index, which contributes to reduce the computation complexity of proposed method. Finally, we improved the deep learning network to detect the targets in various behaviors after the clutter suppression. In order to address the issue of limited clutter data, transfer learning based on historical radar echo data is applied. A thorough experimental analysis on various real clutter data shows that the proposed method performs competitively. Our method shows a superior performance over many state-of-the-art clutter suppression methods, with a covered target detection accuracy of 92.4% under strong spread-clutter background.