Target Identification Method on Ship Formation Imaging for HFSWR

Ship detection has been an area of focus for high-frequency surface wave radar (HFSWR). The detection and identification of ship formation have proven significant in early warning, while studies on the formation identification are limited under the complex background and low resolution of HFSWR. In this paper, we first analyze the motion of HFSWR ship formation, and establish a matching ship formation spatial distribution model. There are two types of formations with rigid structures that are range-Doppler (RD) imaging and studied. Then, we propose a cascade identification method of target on ship formation imaging. The proposed algorithm includes a two-stage formation identification stage. In the first-stage, the extremum detector based on the gray value is employed to find cells with suspicious targets, and the Seed-Filling (SF) algorithm is introduced to connect multi-cell regions belonging to the same ship formation. An extremum detector based on connected regions is proposed to avoid duplicate detection of the same ship formation, and reduce data to be processed in the second-stage. In the second-stage, a lightweight CNN is designed to extract the convolutional features of suspicious targets. The highly abstract features extracted by CNN are input into the extreme learning machine (ELM) for efficient classification. We propose CNN-ELM to identify two densely distributed ship formations from inhomogeneous clutter and single-ship targets. The experimental results based on factual HFSWR background demonstrate that our CNN-ELM model surpasses classic Alexnet and Resnet18 in terms of detection accuracy (97.5%) and processing time (0.871 s). Our proposed cascade identification algorithm is suitable for ship formation identification.