Ship detection in synthetic aperture radar (SAR) images is critical to ocean surveillance and rescue. Although many deep learning SAR ship detection methods have been proposed, the performance of these methods depends on the size and quality of the training samples. To resolve these issues, this letter presents an unsupervised ship detection method in SAR images using superpixel segmentation and cross stage partial network (CSPNet). First, the SAR image is over-segmented into superpixels based on our previously proposed superpixel generation algorithm. Then, the complex signal kurtosis (CSK) and a local superpixel contrast are integrated as a statistical indicator for the automatic identification of ship superpixels and background superpixels, thus leading to generation of training samples. Finally, the segmented superpixels are input to the CSPNet, which can learn a representative feature set with high discrimination ability between ships and backgrounds. Our method can achieve pixel-level detection map rather than the bounding box result. Experiments based on the Gaofen-3 and TerraSAR SAR data demonstrate that our method can achieve above 90% actual detection rate.

The holographic subsurface radar (HSR) is recognized as an effective remote sensing modality for the detection of shallowly buried objects with a high-resolution image in plain view. However, subsurface detection with HSR is prone to be impaired by clutter contamination, which often obscures the target response. In this letter, a novel clutter mitigation method combining singular value decomposition (SVD) and response cross correlation analysis is presented. The proposed method first applies SVD to decompose the radar data matrix to a number of singular components. Furthermore, the signal cross correlation characteristics are analyzed to demonstrate that the variance of left singular vectors is directly proportional to the target proportion in radar data. Then, target and clutter subspaces can be identified by maximizing the defined weighted target-to-clutter ratio (WTCR). Results of numerical simulation and laboratory experiments corroborate the effectiveness of the proposed method in reducing clutter while preserving the target image.

Holographic subsurface radar (HSR) is a promising geophysical electromagnetic technique to detect shallowly buried objects due to its high lateral resolution. However, the subsurface inspections and visualization of buried targets are prone to be impaired by strong clutter from the nonplanar surface reflections. In this article, a clutter mitigation method using dual-frequency cancellation and sparse feature enhancement is proposed for HSR data to distinguish the targets from background. The radar signals are received at two distinct frequencies under certain conditions to calculate the strong surface reflections. After cancellation of the estimated surface, a modified$\ell _{1}$regularization is utilized to further mitigate the residual clutter and highlight the target signature. The effectiveness of the proposed method is evaluated on both numerical simulation and radar signals collected from real HSR systems. The visual and quantitative results demonstrate that the proposed method successfully removes the nonplanar surface clutter with the targets preserved.

The holographic subsurface radar (HSR) has been a promising geophysical electromagnetic technique for detecting shallowly buried targets with high lateral resolution image. However, the radar images are considerably interpreted by strong reflections from the rough surface and inhomogeneity in media of interest. In this article, we focus on mitigating the clutter in HSR applications using a learning-based approach, which requires neither prior information regarding the penetrable medium characteristics nor analytic framework to describe the through-medium interference. The generative adversarial network (GAN) with attentive subspace projection is developed to remove the clutter and recover the target image. The subspaces containing target response are selected with the multi-head attention preliminarily. Then, the generative network will further focus on the target regions and the discriminative network will assess the generated results locally and globally. Experiments using real data were conducted to demonstrate the effectiveness of our approach. The visual and quantitative results show that the proposed approach achieves superior performance on removing clutter in HSR images compared with the state-of-the-art clutter mitigation approaches.

The interrupted-sampling repeater jamming (ISRJ) is coherent and has the characteristic of suppression and deception to degrade the synthetic aperture radar (SAR) image quality. The anti-ISRJ methods are studied in this work in order to suppress the ISRJ based on the waveform and filter design for SAR. First, the relationship between the ISRJ and waveform is obtained by analyzing the principle of the ISRJ using the ambiguity function. The ISRJ produces multiple false targets based on the high Doppler tolerance of waveform and the characteristic of the matched filter. Next, a method is proposed to counter the ISRJ by transmitting a phase-coded (PC) waveform with low Doppler tolerance and designing the corresponding mismatched filter. The joint design method is then developed to improve the anti-ISRJ and imaging performance. In the proposed methods, the majorization minimization framework is introduced to solve the nonconvex waveform and filter design problem. Finally, several simulations are conducted to demonstrate the effectiveness of the proposed methods. Simulation results show that the joint design method shows better anti-ISRJ and imaging performance in comparison with the separate design method, but it is more sensitive to the ISRJ sampling duty ratio and period.

A joint waveform and filter design method is developed for suppressing radar pulse compression sidelobe level in this article. The problem is formulated as the minimization of the integrated sidelobe level (ISL) under the constraint of waveform constant modular and filter energy. To control the loss-in-processing gain (LPG), an additional function is introduced to constrain the peak level based on penalty function method. The joint design algorithm is then derived based on the alternating minimization and majorization minimization (MM) schemes. The computation complexity is analyzed and the convergence analysis verifies that the algorithm can converge to a critical point. Specifically, the proposed method is extended to the waveform and filter design for suppressing the peak sidelobe level (PSL). Numerical simulations are carried out to analyze the key parameters, which demonstrate the feasibility of the proposed method. Simulation results show that the ISL and PSL can be significantly reduced with small LPG. Moreover, the proposed method exhibits faster running speed than the existing one and it thus can be applied to longer sequence designs.

In this letter, a method is proposed to suppress the interrupted sampling repeater jamming (ISRJ) by jointly designing the radar waveform and mismatch filter. The joint design problem under multiple constraints is formulated with the optimization criterion of minimizing the integrated sidelobe levels (ISLs) of waveform mismatch filter output and integrated levels (ILs) of ISRJ signal mismatch filter output. An iterative algorithm is proposed to optimize the waveform and mismatch filter using Lagrange multiplier method and alternating direction multiplier method (ADMM), respectively. Simulation results demonstrate that the proposed method achieves good pulse compression performance while suppressing the ISRJ.

This article proposes an efficient and adaptive statistical superpixel merging approach with edge penalty for polarimetric synthetic aperture radar (PolSAR) image segmentation. Based on the initial superpixel over-segmentation result obtained by our previously proposed adaptive polarimetric superpixel generation algorithm (Pol-ASLIC), this work achieves efficient and accurate PolSAR image segmentation by merging superpixels using the statistical region merging (SRM) framework. This article proposes to define a new dissimilarity measure between superpixels, which takes the edge penalty into consideration, leading to a reasonable and accurate merging order for superpixel pairs. With regard to the merging predicate of superpixels, a polarimetric homogeneity measurement (HoM) is used to define the merging threshold, making the merging predicate and merging threshold adaptive to the PolSAR image content. Experimental results on three airborne and one spaceborne PolSAR data sets demonstrate that the proposed approach can effectively improve the computation efficiency and segmentation accuracy in comparison with state-of-the-art merging-based methods for PolSAR data. More importantly, the proposed approach is free of parameters and easy to use.