Deep learning (DL) based synthetic aperture radar (SAR) imagery ship detection is challenged by multiscale ships on the identical SAR image, which inevitably leads to insufficient and low-quality positive samples during training and ultimately degrades detection performance. To address this issue, we propose a Scale-Sensitive Adaptive Sample Allocation Strategy (SSA-SAS) for SAR ship detection. SSA-SAS ranks candidate boxes using a unified score that integrates a scale-sensitive Wasserstein distance (SSWD), a shape cost, and classification confidence. SSWD serves as the core regression metric, enabling adaptive tolerance to positional offsets based on object scale. Meanwhile, the shape cost introduces morphological priors to guide early-stage optimization. These components jointly enhance the quantity and quality of selected positive samples throughout training. Experimental results show that SSA-SAS improves average precision (AP) by up to 2.6% on the high-resolution SAR images dataset for ship detection and instance segmentation (HRSID) dataset and 1.4% on the SAR ship detection dataset (SSDD), while accelerating network convergence by approximately 5.0%.

With the significant advancements in deep learning technology and the substantial improvement in remote sensing image resolution, remote sensing semantic segmentation has garnered widespread attention. Synthetic aperture radar (SAR) and optical images are the primary sources of remote sensing data, offering complementary information. SAR images can capture surface information even under cloud cover and at night, whereas optical images provide higher resolution in clear weather conditions. Deep learning-based feature fusion methods can effectively integrate multisource information to obtain more comprehensive surface data. However, there are significant spatiotemporal differences in multisource information, making it challenging to select and extract the most discriminative features for segmentation tasks. To address this, we propose a lightweight and efficient fusion semantic segmentation network, SOLSTM, which mixes SAR and optical images as inputs and performs cyclic cross-fusion to establish a new network paradigm. To tackle multisource data heterogeneity, we introduce SAR-OPT matching attention, which aggregates multisource image features by adaptively adjusting fusion weights, thereby achieving comprehensive perception of feature channels and contextual information. Additionally, to mitigate the high computational complexity of processing multidimensional data, we introduce the mLSTM block, which employs linear operations to mine global contextual information in fused images, thus reducing computational complexity and enhancing image segmentation performance. Experiments on the WHU-OPT-SAR dataset show that SOLSTM has excellent performance, achieving up to 52.9 mIoU and outperforming single source image segmentation, verifying the effective fusion of OPT-SAR.

Compared with horizontal detection, oriented ship detection provides accurate target localization and refined boundary delineation. However, ship detection in synthetic aperture radar (SAR) imagery faces significant challenges, including complex backgrounds and densely packed targets. To address these problems, we propose a novel network based on convolutional neural networks (CNNs) and graph neural networks (GNNs), named CNN-GNN-aware detector (CGA-Det). CGA-Det includes three innovations: 1) a CNN-GNN encode network (CG-Encode Network) that captures local and global relationships to concentrate on the targets’ area in complex densely populated scenes; 2) an adaptive feature fusion module (AFFM) that dynamically selects and integrates features from multilevels to enhance detection effect; and 3) a spatial-channel awareness head (SCHead) that promotes directional sensitivity by enhancing spatial and channel representation capacity of the detection head. Experiments on the SAR ship detection dataset (RSSDD) and RSDD-SAR (RSDD) demonstrate the state-of-the-art performance of CGA-Det, with 99.19% and 97.20% mAP50, excelling in complex scenes.

Radar transmits active agile waveformplays a significant role in anti-jamming. However, efficient coherent integration of target's agile echo in a coherent processing interval (CPI) usually poses a severe challenge. This letter proposes an improved sparsity adaptive matching pursue (ISAMP) to address this issue. Firstly, the agile echo signal model of random interpulse frequency and PRT joint agile (RI-FPrtJA) waveform is derived; The sparse reconstruction model of RI-FPrtJA echo signal is then mathematically deduced based on compress sensing theory; Lastly, the ISAMP is proposed to accurately accomplish sparse reconstruction based on the regularized grids mismatch correction and adaptive searching step extension. The simulation results demonstrate that the ISAMP method can achieve better coherent integration in terms of mismatch sidelode levels suppression, sparse reconstruction capacity, and computational cost, compared to some current existing methods.

As a key technology in radar reconnaissance systems, radar signal sorting aims to separate multiple radar pulses from an interleaved pulse stream. Supervised signal sorting methods based on deep learning depend on a large volume of training data to optimize model parameters. However, acquiring labeled pulses in practice is challenging. In this letter, a semi-supervised learning (SSL) framework is proposed to address this issue. First, a Self-Organizing Map (SOM) is used to learn the spatial distribution of impulse features, and an anchor graph is constructed based on SOM nodes. A pseudo-label set is then generated using the SOM based on pulse discrepancy information. Finally, a three-layer Weighted Residual Graph Convolutional Network (WRGCN) is designed for signal sorting, with its parameters pre-trained on pseudo-labels and fine-tuned with a limited number of true labels. Experiments on a simulated radar pulse dataset demonstrate that this framework outperforms several existing methods for radar signal sorting with limited labeled pulses.

Detecting ships in Synthetic Aperture Radar (SAR) images is a challenging task due to various factors, such as the diverse distribution of ships and the intricate nature of SAR images. In recent years, deep learning has made excellent progress in the field of SAR interpretation. Models that focus on extracting global semantic information can effectively achieve balanced detection of multi-scale SAR targets, but their computational complexity is relatively high. Models that focus on processing local information have redundant calculations and poor robustness, but are prone to mistaking the background information of SAR images for targets. To address the above issues, we propose a real-time SAR ship detector based on global semantic information enhancement and local gradient information mining. The lightweight feature extraction backbone based on linear computing is designed, with the network structure of Global Information Augmentation Encoder (GIAE)—Local Gradient Information Miner (LGIM)—Decoder, which can quickly perform feature extraction. GIAE enhances the expression of image content through the long sequence modeling capability of the State Space Model. LGIM uses gradient modules composed of depthwise separable convolutions to extract local information of image, and utilizes directed self-attention (DSA) to mine channel context information. GLDet can complete object detection, rotated object detection and instance segmentation tasks by transforming the detection head. Excellent performance has been achieved on the SAR ship instance segmentation dataset SSDD and HRSID, as well as the SAR rotated ship dataset RSDD-SAR and SSDD+. Meanwhile, GLDet demonstrated excellent generalization performance in large-scale SAR images captured by GF-3 and Terra-SAR satellites.

With the rapid development of Synthetic Aperture Radar (SAR), the number and resolution of SAR images are constantly increasing. As a high-value target, aircraft detection has become a research hotspot in the field of SAR image interpretation. SAR aircraft have diverse postures, complex backgrounds, and small differences among different types of aircraft, which can easily lead to false detections. Meanwhile, some SAR aircraft have incomplete structures and are accompanied by speckle noise, which can easily lead to missed detections. To address the above issues, we propose an ultra-lightweight SAR aircraft detection network based on multi-scale matching attention (MSMANET). Firstly, we propose an ultra-lightweight backbone that extracts SAR gradient features through parallel processing of traditional convolution and Ghost modules. Secondly, aiming to the scale, shape and background information of aircraft, Multi-Scale Matching Attention (MSMA) is designed. MSMA performs feature aggregation and cross channel feature matching on multi receptive field feature maps, making the network more focused on feature maps suitable for detection. The mean average precision (mAP) of MSMANET on the SAR-AIRcraft1.0 dataset is as high as 98.4%, with the 1.6 GFLOPS, 657K parameter and 55.1 FPS. Compared to existing advanced networks, the performance has reached SOTA.

Synthetic Aperture Radar (SAR) plays a crucial role in maritime search, rescue operations and port vessel traffic monitoring. Existing algorithms are difficult to simultaneously extract features of multi-scale targets in SAR images, resulting in uneven accuracy in multi-scale ship instance segmentation. Moreover, due to the complexity of image scenes, existing algorithms struggles to accurately segment targets. Regarding the above difficulties, we propose a Pixel level SAR ship instance segmentation network based on global and local feature cross attention (SSGL). We propose a context aware convolutional attention module (CACA). CACA leverages cross-correlation calculations for global information, aiding SSGL in better distinguishing between foreground objects and complex backgrounds. We design a channel optimization module (COM) that combines multi-path convolution with channel attention to adaptively adjust the receptive field, allowing for balanced feature extraction across different target scales. SSGL’s instance segmentation mask achieved 93.6 on the SSDD dataset and 89.1 on the HRSID dataset. Both APMand APLsignificantly surpasses comparative algorithms, proving SSGL's balanced and high-precision instance segmentation for multi-scale targets.

Deep learning has been extensively applied in SAR ship detection because of its powerful feature extraction ability. However, most methods are not only complex, but also easily lead to missed and false detections when the ships are arranged densely. To meet above challenges, a lightweight network LDSS-Net for dense SAR ship detection is proposed. Firstly, we improve the CSP structure and design a lightweight gradient shunt aggregation backbone network LGSA for better feature extraction while reducing computational overhead. Secondly, a feature fusion network DCE-PAN is proposed for enhancing dense ship contour, which enriches the frequency domain information and improves the local feature correlation by using DWT and ECA. Experiments on public datasets SSDD and HRSID demonstrate that the mAP of LDSS-Net reaches 98.20% and 91.29%, respectively, and the parameters are only 2.0M. Our network outperforms existing advanced networks and achieves excellent detection results.

In modern electromagnetic countermeasure environments, traditional radar signal sorting (RSS) methods face challenges from incompletely intercepted parameter-dense pulses of multi-function radars (MFRs). To cope with this situation, this letter proposes a sequence-to-sequence RSS method based on a multiple self-attention coupling mechanism Transformer network. The method utilizes positional encoding to obtain stable temporal information. A multiple self-attention coupling mechanism is then designed to calculate the attention matrix, thereby extracting sequence relationships for the non-ideal pulse stream. Finally, a decoder network is employed to extract high-dimensional features and translate the corresponding labels for each pulse. Simulation experiments demonstrate that compared with some existing methods, the proposed method can achieve better average sorting accuracy with little computational cost under the conditions of overlapping parameters, limited label, missing pulses, and various modulation types of intercepted MFR signals.

Deep learning has made rapid progress in the field of synthetic aperture radar (SAR) detection. However, SAR images themselves have limited information, and a general detection network that is too wide and too deep can result in computational complexity and memory waste. Therefore, we design a lightweight network for multi-class SAR detection based on adaptive scale distribution attention. Firstly, a novel backbone is designed from the perspective of lightweight model, using deep separable convolution to generate high-quality feature maps of protruding targets, and applying channel shuffle to improve training and detection efficiency. Secondly, a lightweight adaptive scale distribution attention is proposed, which can adaptively obtain the scattering information of multi-scale targets, aggregate the position and contour features of the targets, and improve the detection accuracy of multi-class targets. Finally, anchor-free detection head is applied to improve the generalization ability and robustness of the model. MLSDNet achieve a high mean average precision (mAP) of 92.99% on the newly released multi-class SAR target datasets (MSAR-1.0) with only 1.42G FLOPs and 928.25K Params. The mAP on SAR ship datasets such as SSDD and HRSID reached 99.1% and 94.7%, respectively. The mAP on the latest SAR aircraft dataset reached 97.7%, demonstrating its good generalization ability. Its performance has reached the state-of-the-art (SOTA).

The radon Fourier transform (RFT)-based coherent integration is an important target detection method. However, the computational cost of parameter searching and blind-speed sidelobe (BSSL) remain the main challenges in actual RFT applications. In this letter, we propose a Whale optimization algorithm-based RFT (WOA-RFT) to accelerate the parameter searching process and improve BSSL suppression performance. First, the discrete RFT signal model is derived; WOA-RFT is then proposed to speed up the RFT and suppress BSSL. The simulation results demonstrate that our proposed method has a better performance in terms of parameter estimation accuracy, BSSL suppression capability, and computational cost, compared to some existing methods.

The dense, complex and variable electromagnetic environment poses a serious challenge to radar signal sorting (RSS) in modern electronic reconnaissance systems. In order to improve RSS performance, this letter proposes a semi-supervised learning framework-based RSS method via a residual graph convolutional network (ResGCN-RSS) to effectively improve the generalization ability of the signal sorting models in small data scenarios. Firstly, the graph structure construction of intercepted radar signals is performed via K-nearest neighbor algorithm. Then, the three-layer ResGCN is designed to adaptively improve the features learning. Finally, RSS can be effectively and efficiently implemented through an end-to-end ResGCN with small labeled graph data of interleaved radar signals. The simulation experimental results show that our proposed method can achieve better average accuracy with little computational cost increasing when the labeled data is very small, compared to some existing methods.

In article[1], after drawing the curve of Fig. 4, we mistakenly marked two coordinate values in the fourth column of the figure to be the same as those in the third column. We modified the incorrectly marked coordinates 0.9585 and 0.9572 to the correct values 0.9489 and 0.9483, respectively. The error here is only a marking error; there is no change in the curve of the figure, and it does not affect the discussion and conclusion of this article or any other places. Since the article is early accessed in IEEE, we hope to correct it in the upcoming issue, thank you. The revised figure is shown below:

Speckle noise is an important factor affecting the accuracy of synthetic aperture radar (SAR) target recognition. Traditional speckle reduction methods based on transform domain and spatial filtering usually require professional experience to set the threshold, which will also affect the recognition accuracy. This letter proposes a multilevel wavelet speckle reduction network (Wavelet-SRNet) for noisy SAR images target recognition. First, the method designs the wavelet soft threshold denoising method as a trainable neural network module in the convolutional neural network (CNN) framework. Then, a two-level wavelet denoising branch is constructed and fused with the original noisy image. Finally, we cascade a CNN-based classification model on the above structure to form an SAR image target recognition network whose denoising threshold can be automatically learned. Experiments on the moving and stationary target acquisition and recognition database show that the classification accuracy of the proposed method for target recognition in noisy SAR images is better than the compared state-of-the-art methods. Also, the method achieved high test accuracy in the noise augmentation experiment.

Radar signal sorting (RSS) plays an important role in the electronic support measurement system. However, the existing clustering-based RSS methods depend heavily on prior knowledge to achieve excellent performance, which may bring severe challenges to RSS in actual scenarios. This letter proposes a novel subspace decomposition based adaptive density peak clustering (SD-ADPC) method to address the problems of low accuracy and high computational cost in RSS. First, the original complex radar signal data is directly decomposed into two-dimensional (2D) subspace by t-distributed random neighborhood embedding (t-SNE). Then, based on the outlier detection of the products between the peak density and the distance of the data points, ADPC is used to adaptively determine the optimal clustering centers of the original data in 2D subspace. Finally, the reminding data is assigned to its nearest cluster with Euclidean distance in one step. The experimental results of the simulated RSS dataset and the open baselines show that our proposed method does not require any knowledge and can achieve better or competitive performance in terms of accuracy and computational cost, compared to existing state-of-the-art methods.