Shipborne HFSWR Target Detection in Sea Clutter Regions Based on 3-D Feature Fusion

Abstract

Shipborne high-frequency surface wave radar (HFSWR) systems face the challenge of sea clutter spreading, which obscures vessel echoes and makes detection difficult. In this article, we propose a novel 3-D target detection algorithm that effectively identifies vessel targets in sea clutter using multidimensional fusion features. The algorithm consists of two stages: 3-D spectrum construction and target detection. In the 3-D spectrum construction stage, the digital narrow beam forming (DNBF) method is combined to transform the range-Doppler (RD) spectrum into a range-Doppler–azimuth 3-D spectrum. In the target detection stage, a two-level cascade target detection algorithm is proposed. At the first level, a 3-D extremum detection algorithm identifies potential vessels in sea clutter from the 3-D spectrum and locates the 3-D tensor blocks containing high-dimensional morphology features of these potential vessels. At the second level, we introduce an intelligent 3-D tensor block classifier, which includes a two-channel 3-D feature-extraction network and a feature classifier. This network extracts 3-D morphology features from the tensor blocks using 3-D discrete wavelet transform and a 3-D convolutional neural network (CNN). The extracted features are then fused using robust sparse linear discriminant analysis (RSLDA), and an extreme learning machine processes the fusion features to produce the final results. The experimental results show that the proposed algorithm outperforms state-of-the-art methods in terms of detection rate and false alarm rate.