Abstract—In recent years, convolutional neural networks (CNNs) have drawn considerable attention for the analysis of synthetic aperture radar (SAR) data. In this study, major subareas of SAR data analysis that have been tackled by CNNs are systematically reviewed, such as automatic target recognition, land use and land cover classification, segmentation, change detection, object detection and image denoising. Special emphasis has been given to practical techniques such as data augmentation and transfer learning. Complex-valued CNNs, which have been introduced to exploit phase information embedded in SAR complex images, have also been extensively reviewed. To conclude this review paper, open challenges and future research directions are highlighted.

SAR image targets detection and recognition (SAR-TDR) have become research hotpots in the remote sensing application. These targets include ships, vehicles, aircrafts, oil tanks, bridges, and so on. However, with the rapid development of SAR technology and increasingly complex electromagnetic environment, complex characteristics of SAR images bring severe challenges to the accurate SAR-TDR via traditional physical models or manually features extraction based machine learning methods. In recent years, deep learning (DL), as a powerful automatic features extraction algorithm, has been widely used in the computer vision domain. More specifically, DL has also been introduced into the SAR-TDR tasks and effectively achieved good performances in terms of accuracy, real-time, etc. With the rapid development of DL, SAR images processing, and practical requirements of SAR-TDR in civilian and military domains, it is very necessary to make a systematic survey on the SAR-TDR in past few years. In this survey paper, we mainly conduct a systematic overview of DL-based SAR-TDR literatures on two tasks, i.e., targets recognition (e.g., ground vehicles, ships, and aircrafts) and targets detection (e.g., ships, aircrafts,, change detection, seasurface oil spill, and oil tanks), respectively. More specifically, our related works about these topics are also presented to verify the effectiveness of DL-based methods. Firstly, several DL methods (e.g., CNN, RNN, AEs, and GNN), commonly used to SAR-TDR, are briefly introduced. The systematic review of DL-based SAR-TDR (including our related works) is then presented. Finally, the current challenges and future possible research directions are deeply analyzed and discussed.

The applicability of compressive sensing (CS) to inverse synthetic aperture radar (ISAR) imagery has been widely discussed over the past few years. In particular, CS-based ISAR image-reconstruction algorithms have been developed and their effectiveness proven when dealing with incomplete ISAR data. Resolution enhancement has also been identified as a case for which CS can be effectively applied to ISAR imagery. In this case, the acquired signal can be interpreted as incomplete data in the frequency/slow-time domain and CS used to reconstruct the super-resolved ISAR image. In this paper, an exhaustive performance analysis is carried out along with a comparison between CS and conventional super-resolution techniques. Several concepts and methods have been introduced in order to effectively define the performance, which is not simply based on visual inspection.

Phase autofocus is a significant step in translational motion compensation for inverse synthetic aperture radar (ISAR) imaging. Among the existing autofocus methods, contrast maximization-based algorithms are superior in accuracy and robustness. However, the existing maximum contrast autofocus methods are parametric, which significantly affects their convergence speed and, in some cases, limits their applicability. In this paper, a novel non-parametric maximum contrast ISAR autofocus algorithm based on Newton's method is proposed to inherit the high level of accuracy and robustness of contrast-based algorithms and, at the same time, achieve faster convergence. The simplified Newton's method, the modified Newton's method, and the secant method are used to obtain a higher computational efficiency. The proposed method is then compared to a well-established non-parametric method, namely the minimum entropy autofocus method. It will be shown that the proposed method can improve the computational efficiency by a factor from 5 to 10 times in typical scenarios. Both simulated and real data will be used to test the proposed algorithm, particularly in terms of computational effectiveness.

Deep-learning-based synthetic aperture radar automatic target recognition (SAR-ATR) plays a significant role in the military and civilian fields. However, data limitation and large computational cost are still severe challenges in the actual application of SAR-ATR. To improve the performance of the convolutional neural network (CNN) model with limited data samples in SAR-ATR, this article proposes a novel multidomain feature subspace fusion representation learning method, i.e., a lightweight cascaded multidomain attention network, namely, LW-CMDANet. First, we design a four-layer CNN model to perform hierarchical feature representation learning via the hinge loss function, which can efficiently alleviate the overfitting problem of the CNN model by a nongreedy training style with a small dataset. Then, a cascaded multidomain attention module, based on discrete cosine transform and discrete wavelet transform, is embedded into the previous CNN to further complete the class-specific feature extraction from both the frequency and wavelet transform domains of the input feature maps. Thus, the multidomain attention can enhance the feature extraction ability of previous nongreedy learning manner, to effectively improve the recognition accuracy of the CNN model. Experimental results on small SAR datasets show that our proposed method can achieve better or competitive performance than that of many current existing state-of-the-art methods in terms of recognition accuracy and computational cost.

Three-dimensional (3D) ISAR imaging has been proven feasible by combining traditional ISAR imaging and interferometry. Such technique, namely Inteferometric ISAR (InISAR), allows for the main target scattering centres to be mapped into a 3D spatial domain, therefore forming 3D images under the form of 3D point clouds. 3D InISAR overcomes some main limitations of traditional 2D ISAR imaging, such as the problem of cross-range scaling and unknown image projection plane (IPP). Despite the great advantage of 3D imaging over traditional 2D imaging, some issues remain, such as scatterer scintillation, shadowing effects, poor SNR, etc., which limit the effectiveness of 3D imaging. A solution to these issues can be found in the use of multiple 3D views, which can be obtained exploiting either multi-temporal or multi-perspective configurations or a combination of both. This paper propose this concept and develops the image fusion algorithms that are necessary to produce multi-view 3D ISAR images. The effectiveness of the proposed technique is tested by using real data collected with a multi-static InISAR system.