Targets In Synthetic Aperture Radar Images Research Articles

Target Detection Network for SAR Images Based on Semi-Supervised Learning and Attention Mechanism

The existing Synthetic Aperture Radar (SAR) image target detection methods based on convolutional neural networks (CNNs) have achieved remarkable performance, but these methods require a large number of target-level labeled training samples to train the network. Moreover, some clutter is very similar to targets in SAR images with complex scenes, making the target detection task very difficult. Therefore, a SAR target detection network based on a semi-supervised learning and attention mechanism is proposed in this paper. Since the image-level label simply marks whether the image contains the target of interest or not, which is easier to be labeled than the target-level label, the proposed method uses a small number of target-level labeled training samples and a large number of image-level labeled training samples to train the network with a semi-supervised learning algorithm. The proposed network consists of a detection branch and a scene recognition branch with a feature extraction module and an attention module shared between these two branches. The feature extraction module can extract the deep features of the input SAR images, and the attention module can guide the network to focus on the target of interest while suppressing the clutter. During the semi-supervised learning process, the target-level labeled training samples will pass through the detection branch, while the image-level labeled training samples will pass through the scene recognition branch. During the test process, considering the help of global scene information in SAR images for detection, a novel coarse-to-fine detection procedure is proposed. After the coarse scene recognition determining whether the input SAR image contains the target of interest or not, the fine target detection is performed on the image that may contain the target. The experimental results based on the measured SAR dataset demonstrate that the proposed method can achieve better performance than the existing methods.

Recognition of occluded targets in SAR images based on matching of attributed scattering centers

ABSTRACT A local optimal matching scheme of attributed scattering centres (ASC) is proposed for occluded target recognition in synthetic aperture radar (SAR) images. Considering the number of matching pairs and differences between different matching pairs, the local optimal matching is established between the ASC sets from the test and template samples. Accordingly, the properties of occluded targets are used to define a robust similarity measure. The similarities between the test sample and different template samples are calculated to determine the target label. Experiments are conducted on the Moving and Stationary Target Acquisition and Recognition (MSTAR) dataset. The proposed method could achieve an average recognition rate of 99.18% for 10 classes of targets under the standard operating condition (SOC). Its performance under random occlusion and directional occlusion is also superior over the reference methods.

SAR ATR for Limited Training Data Using DS-AE Network.

Although automatic target recognition (ATR) with synthetic aperture radar (SAR) images has been one of the most important research topics, there is an inherent problem of performance degradation when the number of labeled SAR target images for training a classifier is limited. To address this problem, this article proposes a double squeeze-adaptive excitation (DS-AE) network where new channel attention modules are inserted into the convolutional neural network (CNN) with a modified ResNet18 architecture. Based on the squeeze-excitation (SE) network that employs a representative channel attention mechanism, the squeeze operation of the DS-AE network is carried out by additional fully connected layers to prevent drastic loss in the original channel information. Then, the subsequent excitation operation is performed by a new activation function, called the parametric sigmoid, to improve the adaptivity of selective emphasis of the useful channel information. Using the public SAR target dataset, the recognition rates from different network structures are compared by reducing the number of training images. The analysis results and performance comparison demonstrate that the DS-AE network showed much more improved SAR target recognition performances for small training datasets in relation to the CNN without channel attention modules and with the conventional SE channel attention modules.

CNN-LRP: Understanding Convolutional Neural Networks Performance for Target Recognition in SAR Images.

Target recognition is one of the most challenging tasks in synthetic aperture radar (SAR) image processing since it is highly affected by a series of pre-processing techniques which usually require sophisticated manipulation for different data and consume huge calculation resources. To alleviate this limitation, numerous deep-learning based target recognition methods are proposed, particularly combined with convolutional neural network (CNN) due to its strong capability of data abstraction and end-to-end structure. In this case, although complex pre-processing can be avoided, the inner mechanism of CNN is still unclear. Such a “black box” only tells a result but not what CNN learned from the input data, thus it is difficult for researchers to further analyze the causes of errors. Layer-wise relevance propagation (LRP) is a prevalent pixel-level rearrangement algorithm to visualize neural networks’ inner mechanism. LRP is usually applied in sparse auto-encoder with only fully-connected layers rather than CNN, but such network structure usually obtains much lower recognition accuracy than CNN. In this paper, we propose a novel LRP algorithm particularly designed for understanding CNN’s performance on SAR image target recognition. We provide a concise form of the correlation between output of a layer and weights of the next layer in CNNs. The proposed method can provide positive and negative contributions in input SAR images for CNN’s classification, viewed as a clear visual understanding of CNN’s recognition mechanism. Numerous experimental results demonstrate the proposed method outperforms common LRP.

Channel Attention Module in Convolutional Neural Network and Its Application to SAR Target Recognition Under Limited Angular Diversity Condition

In the field of automatic target recognition(ATR) with synthetic aperture radar(SAR) imagery, it is usually impractical to obtain SAR target images covering a full range of aspect views. When the database consists of SAR target images with limited angular diversity, it can lead to performance degradation of the SAR-ATR system. To address this problem, this paper proposes a deep learning-based method where channel attention modules(CAMs) are inserted to a convolutional neural network(CNN). Motivated by the idea of the squeeze-and-excitation(SE) network, the CAM is considered to help improve recognition performance by selectively emphasizing discriminative features and suppressing ones with less information. After testing various CAM types included in the ResNet18-type base network, the SE CAM and its modified forms are applied to SAR target recognition using MSTAR dataset with different reduction ratios in order to validate recognition performance improvement under the limited angular diversity condition. Keywords: Synthetic Aperture Radar, Automatic Target Recognition, Convolutional Neural Network, Channel Attention Module Keywords: 합성 개구면 ë ˆì´ë”, 자동 í‘œì  식별, 합성곱 ì‹ ê²½ë§, 채널 특징 집중 모듈

SAR Image Classification Using Contrastive Learning and Pseudo-Labels With Limited Data

Synthetic aperture radar (SAR) image classification based on deep learning methods has attracted huge enthusiasm among researchers. Sufficient training data guarantee the generalization ability of neural networks. However, it is quite difficult to collect large amount of high-quality and labeled SAR data in real situations. In this letter, a method based on contrastive learning and pseudo-labels is proposed to solve the few-shot SAR image classification problem. First, a siamese structure is introduced to learn semantic representations of SAR target images. The feature vectors obtained in this stage can reveal the similarity between bi-temporal SAR images. Then, the backbone network of the siamese network is connected to a classification head for instance-level classification. The pseudo-label strategy is adopted to gradually improve the performance. A loss function that changes with iteration is designed to optimize the training process with true labels and pseudo-labels. Besides, considering the imprecision of single-network architecture, we adopt dual-networks and cross-training strategies. The final result is obtained by integrating the results of the dual-networks. Ablation studies show that the designed components of the proposed method are effective. Comparative experiments on the moving and stationary target acquisition and recognition (MSTAR) data set demonstrate the superior performance of the proposed method when dealing with limited training data in the SAR classification.

Region adaptive morphological reconstruction fuzzy C-means for near-field 3-D SAR image target extraction

Recently, near-field 3-D synthetic aperture radar (SAR) imaging has become a research focus in the SAR field. The target extraction is an important step in 3-D SAR image application. As far as we know, there are few studies on the target extraction for 3-D SAR images. One of the difficulties that the target extraction has to face is the influence of the environmental interference in the background. In this paper, we propose a region adaptive morphological reconstruction fuzzy C-means algorithm to overcome the influence of interference and achieve high-precision target extraction for 3-D SAR images. Firstly, 3-D anisotropic diffusion filtering achieves image smoothing with edge preserved. Secondly, a 3-D omnidirectional detection operator is constructed to obtain the gradient magnitude of the images. The target edges, which provide coarse target position information, are extracted by the hysteresis threshold method. Then, multiple sub-images are extracted from the original image based on the position information. Thereby, the influence of interference is eliminated, and the coarse target region is obtained. Finally, the adaptive morphological reconstruction fuzzy C-means algorithm is utilized to extract the target region more accurately for each sub-image. After the extraction results of all sub-images are merged into one image according to the position information, the high-precision target extraction of 3-D SAR image is completed. For the experiments, we compare the proposed algorithm and 11 algorithms through a real 3-D SAR dataset. The measurement results demonstrate that the proposed algorithm can overcome the influence of interference, and the performance of the proposed algorithm is much higher than that of the other algorithms. Moreover, the proposed algorithm has high computational efficiency.

Resolution Conversion of SAR Target Images Using Conditional GAN

For successful automatic target recognition(ATR) with synthetic aperture radar(SAR) imagery, SAR target images of the database should have the identical or highly similar resolution with those collected from SAR sensors. However, it is time-consuming or infeasible to construct the multiple databases with different resolutions depending on the operating SAR system. In this paper, an approach for resolution conversion of SAR target images is proposed based on conditional generative adversarial network(cGAN). First, a number of pairs consisting of SAR target images with two different resolutions are obtained via SAR simulation and then used to train the cGAN model. Finally, the model generates the SAR target image whose resolution is converted from the original one. The similarity analysis is performed to validate reliability of the generated images. The cGAN model is further applied to measured MSTAR SAR target images in order to estimate its potential for real application. Keywords: 합성 개구면 ë ˆì´ë”, 조건부 ìƒì„±ì  ì ëŒ€ ì‹ ê²½ë§, í‘œì ì˜ìƒ, 해상도 변환 Keywords: Synthetic Aperture Radar, Conditional Generative Adversarial Network, Target Image, Resolution Conversion

Multilevel Wavelet-SRNet for SAR Target Recognition

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.

Fine Building Segmentation in High-Resolution SAR Images Via Selective Pyramid Dilated Network

The building extraction from synthetic aperture radar (SAR) images has always been a challenging research topic. Recently, the deep convolution neural network brings excellent improvements in SAR segmentation. The fully convolutional network and other variants are widely transferred to the SAR studies because of their high precision in optical images. They are still limited by their processing in terms of the geometric distortion of buildings, the variability of building structures, and scattering interference between adjacent targets in the SAR images. In this article, a unified framework called selective spatial pyramid dilated (SSPD) network is proposed for the fine building segmentation in SAR images. First, we propose a novel encoder–decoder structure for the fine building feature reconstruction. The enhanced encoder and the dual-stage decoder, composed of the CBM and the SSPD module, extract and recover the crucial multiscale information better. Second, we design the multilayer SSPD module based on the selective spatial attention. The multiscale building information with different attention on multiple branches is combined, optimized, and adaptively selected for adaptive filtering and extracting features of complex multiscale building targets in SAR images. Third, according to the building features and SAR imaging mechanism, a new loss function called L-shape weighting loss (LWloss) is proposed to heighten the attention on the L-shape footprint characteristics of the buildings and reduce the missing detection of line buildings. Besides, LWloss can also alleviate the class imbalance problem in the optimization stage. Finally, the experiments on a large-scene SAR image dataset demonstrate the effectiveness of the proposed method and verify its superiority over other approaches, such as the region-based Markov random field, U-net, and DeepLabv3+.

A SAR Target Image Simulation Method With DNN Embedded to Calculate Electromagnetic Reflection

Electromagnetic (EM) scattering calculation is a very important part of most synthetic aperture radar (SAR) target image simulation methods. It affects the intensity of the radar echo signal to a great extent, thus affecting the quality of the final simulation image. EM reflection models are usually approximate formulas derived under certain assumptions. The errors between these models and the actual situation can cause significant differences between simulated images and real images. To solve this problem, we propose a novel modified SAR target image simulation framework, in which the deep neural network (DNN) is embedded to calculate the EM reflection, so that the DNN can directly learn and fit the EM reflection models from real SAR images. First, the intensity calculation of radar signal in a single reflection is separated from the cumulative calculation of multiple radar reflection signals intensity in each pixel. Thus, the approximate calculation formulas of EM reflection can be replaced with the DNN models. Next, the DNN model is trained with the backpropagation algorithm to learn the actual EM reflection model from real SAR images. Finally, the fitted EM reflection models and an image post-processing model are applied to simulate images of the target under different imaging angles. In the simulation framework, the functions of the neural network models are limited to calculating the reflection coefficient and adding sidelobe and speckle noise. The imaging model is still the original simulation method based on ray tracing, which ensures the correctness and generalization of the simulation method. Experiments show that the proposed simulation method can significantly improve the quality of the simulation image. When the image is normalized to [0, 1], the minimum mean square error between the simulated SAR images and the real images of the Sandia laboratory implementation of cylinders target can reach 0.003. The visualization results of the DNN models show that the fitted reflection coefficient calculation curve and the convolution kernel used for image post-processing are consistent with the laws in the theoretical model. In addition, when the proposed method is used to simulate complex targets, the similarity of simulation images can also be significantly improved.

A Novel SAR Image Target Recognition Algorithm under Big Data Analysis

Synthetic aperture radar (SAR) image target recognition technology is aimed at automatically determining the presence or absence of target information from the input SAR image and improving the efficiency and accuracy of SAR image interpretation. Based on big data analysis, dirty data is removed, clean data is returned, and standardized processing of SAR image data is realized. At the same time, by establishing a statistical model of coherent speckles, the convolutional autoencoder is used to denoise the SAR image. Finally, the network model modified by softmax cross‐entropy loss and Fisher loss is used for automatic target recognition. Based on the MSTAR data set, two scene graphs containing the target synthesized by the background image and the target slice are used for experiments. Several comparative experiments have verified the effectiveness of the classification and recognition model in this paper.

Synthetic Aperture Radar Image Target Recognition Based on Updated Classifier

Synthetic Aperture Radar Image Target Recognition Based on Updated Classifier

Ship Target Imaging in Airborne SAR System Based on Automatic Image Segmentation and ISAR Technique

The airborne synthetic aperture radar (SAR) image of ship target will be blurred for the complex motion of the target, which will influence the performance of feature extraction and target classification. In this article, a refocusing method of ship target in airborne SAR image though the inverse synthetic aperture radar (ISAR) technique is presented. The novel contributions of this article can be summarized as follows: 1) the residual phase of processed signal is analyzed in detail to illustrate the necessity of adopting ISAR technique; 2) a valid, labor-saving, and automatic image segmentation approach via clustering algorithm is put forward for automatic data extraction and inverse mapping. The clustering algorithm we choose is the agglomerative hierarchical cluster algorithm, which does not require the number of categories in advance; 3) the refocused individual ISAR image of individual ship target is yielded though the range-instantaneous Doppler algorithm. Finally, the simulated and real measured data are processed, and imaging results verify the effectiveness of the novel algorithm presented in this article.

Joint Track-Before-Detect Algorithm for High-Maneuvering Target Indication in Video SAR

Moving target shadows continuously appear in the sequential images produced by video synthetic aperture radar (SAR), which conducts research on shadow-based detection and tracking of moving targets. This article addresses a typical shadow-based detection and tracking problem of dim high-maneuvering targets in the complex background. Building upon the dynamic programming-based track-before-detect (DP-TBD) algorithm, this article proposes a joint-processing-strategy-based DP-TBD (JP-DP-TBD) algorithm to detect and track multiple high-maneuvering targets in video SAR. Benefiting from both of the location and radial velocity information of the targets in the video SAR image and range-Doppler (RD) spectrum, the JP-DP-TBD algorithm can screen the state candidate region and retain more states that conform to the motion law of the target. Meanwhile, we adopt a dual-frame close-range matching mechanism to complete the automatic initialization of the candidate target state, whose validity has been verified on the real video SAR data. Experimental results on simulated data demonstrate that the proposed algorithm has better detection ability and fewer false alarms in comparison with other TBD algorithms.

Multivariate empirical mode decomposition with application to SAR image target recognition

A synthetic aperture radar (SAR) target recognition method was proposed based on multivariate empirical mode decomposition (MEMD). MEMD was the general extension of traditional EMD, which could avoid the mode mixing problems. MEMD was employed to process SAR images to obtain the multi-layer intrinsic mode functions (IMF), which could better reflect the time-frequency properties of the targets. Different layers of IMFs could effectively complement each other while sharing some inner correlations because they are generated from the same target. In the classification phase, the joint sparse representation was employed to represent the IMFs. The joint sparse representation could solve several related sparse representation tasks based on the idea of multi-task learning. It could produce more precise estimations than the solutions of single tasks. According to the sparse coefficient vectors corresponding to different IMFs, the reconstruction errors of different classes for the representation of the test sample can be calculated. Afterwards, the target label of the test sample can be determined. Experiments were conducted on the Moving and Stationary Target Acquisition and Recognition (MSTAR) dataset, by comparison with existing methods under the standard operating condition, depression angle variance, noise corruption, and target occlusion, the results confirm the validity of the proposed method.

Two-stage ship detection in synthetic aperture radar images based on attention mechanism and extended pooling

Object detection in synthetic aperture radar (SAR) images remains a challenging problem due to the particular imaging mechanism of SAR systems. The sizes of targets are relatively small and the scenes are large, indicating that the intersection over union value between the targets and anchors is probably small. In addition, SAR images are severely polluted by speckles under normal conditions. The edges of objects in SAR images are blurred. In this study, we proposed an improved object-detection framework based on a two-stage faster region-based convolutional neural network. A feature-enhancement module based on self-attention mechanism is designed to learn a target-oriented feature map, where the spatial attention and channel attention work simultaneously, such that targets can be highlighted and the speckles can be suppressed to a certain extent. Two separate feature maps serve as the input of the region proposal network to isolate the classification and regression tasks. Because only one category exists in the ship-detection task, correctly distinguishing targets from the background results in the correct detection of ships. In the final classification stage, an extended region-of-interest pooling operation is performed on the potential proposals and contexture information. The usage of extra information can improve target fine-tuning in the final network. To avoid ignoring small targets, we carefully set the anchors’ parameters based on the analysis of ground truth and select the appropriate shapes of feature maps. With the help of these modifications, the proposed method can detect small, weak, and dense targets in SAR images. Ablation experiments over networks with different configurations prove that the proposed modules are working. Experiments on real SAR Gaofen-3 and Sentinel-1 images demonstrate the efficiency of the proposed object-detection framework.

L(1/2) Regularization for ISAR Imaging and Target Enhancement of Complex Image

Synthetic aperture radar (SAR) and inverse synthetic aperture radar (ISAR) imaging technology are powerful tools to acquire high-resolution radar image, which is an important basis for further automatic target recognition (ATR). For ISAR, if the radar frequency is high enough, signals that the radar received usually have strong sparsity when the radar data convert to Fourier domain, and they can be downsampled and restored by compressed sensing (CS). As for the regularization method in CS theory, while L₁ regularization is popular, the Lq(0 < q < 1) regularization is proved to be a sparse regularization framework and could achieve well performance. In this work, inspired by the L₁-based complex approximated message passing (CAMP) method and the L(1/2) regularization framework, we extend the CAMP method into an L(1/2)-based iterative thresholding method in ISAR imaging under the downsampling rate of the scattered fields. The matched filtering (MF) method is widely used to generate the SAR image, where targets are usually overwhelmed by noise and scene background. In order to make the target enhanced to further improve recognition, regularization is adopted in recovering sparse solution and the clutter reduction of MF image. This work implements the L(1/2)-based regularization into the SAR image target enhancement and the scene-noise reduction via CAMP. The SAR image is generated by ideal point scatterers and the real measurement data of RADARSAT-1. Given the sparsity estimated, the experiment results show the advantage of L(1/2) regularization compared with the L₁ regularization.

SAR Target Classification Based on Sample Spectral Regularization

Synthetic Aperture Radar (SAR) target classification is an important branch of SAR image interpretation. The deep learning based SAR target classification algorithms have made remarkable achievements. But the acquisition and annotation of SAR target images are time-consuming and laborious, and it is difficult to obtain sufficient training data in many cases. The insufficient training data can make deep learning based models suffering from over-fitting, which will severely limit their wide application in SAR target classification. Motivated by the above problem, this paper employs transfer-learning to transfer the prior knowledge learned from a simulated SAR dataset to a real SAR dataset. To overcome the sample restriction problem caused by the poor feature discriminability for real SAR data. A simple and effective sample spectral regularization method is proposed, which can regularize the singular values of each SAR image feature to improve the feature discriminability. Based on the proposed regularization method, we design a transfer-learning pipeline to leverage the simulated SAR data as well as acquire better feature discriminability. The experimental results indicate that the proposed method is feasible for the sample restriction problem in SAR target classification. Furthermore, the proposed method can improve the classification accuracy when relatively sufficient training data is available, and it can be plugged into any convolutional neural network (CNN) based SAR classification models.

SAR multi‐target interactive motion recognition based on convolutional neural networks

Synthetic aperture radar (SAR) multi-target interactive motion recognition classifies the type of interactive motion and generates descriptions of the interactive motions at the semantic level by considering the relevance of multi-target motions. A method for SAR multi-target interactive motion recognition is proposed, which includes moving target detection, target type recognition, interactive motion feature extraction, and multi-target interactive motion type recognition. Wavelet thresholding denoising combined with a convolutional neural network (CNN) is proposed for target type recognition. The method performs wavelet thresholding denoising on SAR target images and then uses an eight-layer CNN named EilNet to achieve target recognition. After target type recognition, a multi-target interactive motion type recognition method is proposed. A motion feature matrix is constructed for recognition and a four-layer CNN named FolNet is designed to perform interactive motion type recognition. A motion simulation dataset based on the MSTAR dataset is built, which includes four kinds of interactive motions by two moving targets. The experimental results show that the recognition performance of the authors' Wavelet + EilNet method for target type recognition and FolNet for multi-target interactive motion type recognition are both better than other methods. Thus, the proposed method is an effective method for SAR multi-target interactive motion recognition.