Synthetic aperture radar (SAR) automatic target recognition (ATR) has been widely used by scholars around the world and achieved excellent results. However, occluded SAR target recognition is still a very challenging task. In this letter, we propose a center local constraint shadow residual network (ClcsrNet) for occluded SAR target recognition. First, the shadow features of SAR images are extracted to improve the robustness of the network to occlusion. Then, the shadow features, the target convolutional features, and the residual features are fused to increase the feature diversity of the network. Finally, we combine the center loss and the local constraint loss to optimize the network. The center loss is used to better cluster the targets in the same class. The local constraint loss is used to maintain the local structure of the target, which increases the separability between different classes. Experiments on the moving and stationary target acquisition and recognition (MSTAR) datasets demonstrate that the proposed ClcsrNet can achieve higher accuracy and better robustness than the comparison algorithms in occluded SAR target recognition.

Convolutional neural networks (CNNs) have been widely employed in synthetic aperture radar (SAR) target recognition due to their powerful feature extraction capability. However, the performance of CNN-based SAR target recognition algorithms is often affected by imbalanced datasets, in which some classes own plenty of samples and some classes own few samples. To address this issue, this letter proposes a large-margin Gaussian mixture with a ring loss network (LGM-RNet). To improve CNN’s recognition performance for classes with few samples, the algorithm clusters features of each class in the feature space and makes all the data to be equally distributed on a circle. Furthermore, to mitigate the impact of speckle noise in SAR images on target recognition, a denoising method based on Euclidean loss and the total variation loss is introduced. The proposed algorithm aims to improve the accuracy and robustness of imbalanced SAR image target recognition. Experimental results have verified the effectiveness of the proposed algorithm.

Synthetic aperture radar (SAR) is a critical imaging technique that is widely used for civil and military tasks, as it is featured with an excellent ability for high resolution imaging. However, due to the severe shortage of SAR images, the performance of automatic target recognition (ATR) is greatly sabotaged. Generative adversarial network (GAN) is often applied for data augmentation of small-sized dataset. In this paper, based on auxiliary classifier GAN (ACGAN) and top-k training technique, we propose double top-k training, which implements a modification during training without any further adjustment on model architecture. The proposed method is to enforce generator to only optimize on generated images that perform well in both discriminator and auxiliary classifier, and discard images of poor performance. We evaluate the generated images via recognition on the moving and stationary target acquisition and recognition (MSTAR) dataset. Recognition accuracy and Fréchet inception distance (FID) score indicate better generation results of the proposed method compared with original ACGAN.

Superpixel can maintain the boundary of the target and reduce the influence of speckle noise, which has been widely applied to synthetic aperture radar (SAR) image target detection. But the size of the superpixel has a great impact on the performance of superpixel-based SAR target detection algorithms. To solve this problem, we propose a multi-level ship target detection algorithm based on superpixel segmentation. Firstly, the SAR images are segmented in different levels with different superpixel sizes. Different descriptions of the SAR images are obtained in different levels. Secondly, we determine the feature of the superpixels in each level. And in order to enhance the adaptability of the proposed algorithm, we propose an adaptive distance calculation method to select the contrast superpixels in each level. Thirdly, the soft detection results are realized in each level by using the fuzzy C-means (FCM) algorithm. At last, the soft detection results obtained in different levels are fused by a new fusion strategy to achieve the final ship target detection result. The influences caused by different superxiel sizes can be effectively eased by fusion. Experiments in different SAR images have verified the effectiveness of the proposed algorithm in accurately detecting ship targets and insensitivity to the superpixel size.

In this paper, a full-resolution image segmentation model with multi-source input information is proposed and applied to road extraction. The convolution-deconvolution network is adopted as the backbone network, and a full resolution network branch is added into the backbone network. A data exchange mechanism is established between the backbone network and the full resolution branch, which not only overcomes the problems of reduced feature resolution and loss of detailed information caused by repeated pooling operations, but also aggregates multi-scale features in convolution stage. The aggregated features are transferred to the corresponding layers in deconvolution stage, which enhances the feature fusion. Multi-source images are used as input, and the predictions are fused by weighting at the end of the network to highlight the target while effectively suppressing the misclassification. Experiments on Road Detection Dataset show that the results of the proposed method are superior to those of state-of-the-art comparison methods.

Focusing on the problem of target recognition in complex land backgrounds, a target identification method using scanning correlation and beam pointing is proposed in this paper. Considering the complex scattering characteristics of ground clutters, correlation thresholds associated with range and angles are set to eliminate false targets. Firstly, seeker scanning is implemented twice from left to right, then from right to left to eliminate the negative influences of ground clutters. The information of the detected targets during scanning including range and angles are stored. Point or body identification is realized based on the high-resolution range profile (HRRP). Finally, error distribution area is calculated by combining with the beam pointing information from the fire control system. Target identification is realized by finding the nearest target.

Synthetic aperture radar (SAR) target configuration recognition is a challenging task, and the key point is to realize effective feature extraction. An algorithm combing the advantages of sparse graph and locality preserving projections (LPP) is proposed to achieve SAR target configuration recognition. Taking the merits of sparse representation (SR) into consideration, an affinity matrix is established to realize effective structure preserving of the dataset. Besides, the problem of matrix singularity in LPP is effectively resolved by diagonal loading. Experimental results on the moving and stationary target acquisition and recognition (MSTAR) database validate the effectiveness and superiority of the proposed algorithm.

Focusing on the problem of the real-time implementation in sparse representation (SR) based recognition algorithm, a fast sparse representation (FSR) algorithm is presented in this paper to improve the efficiency of synthetic aperture radar (SAR) target configuration recognition. Taking the inertia variance characteristic of SAR target images over a small range of azimuth angles into consideration, training samples of each configuration are averaged. Instead of using all the training samples to establish the dictionary in SR, the average samples are utilized to construct the dictionary in FSR. A small dictionary accelerates the speed of the proposed algorithm.

It is very difficult for the millimeter wave (MMW) seeker to detect and capture the target. Tracking the target unstably, even losing the target happens frequently. Focusing on the problem, a simple but effective target recapture method is presented for narrow-beam MMW seeker in this paper. The parameters outputted by the inertial navigation system (INS) and the seeker are utilized to deduce the coordinates of the target. And then, target searching is implemented again on the basis of the deduced coordinates. The target recapture time can be dramatically reduced by using the proposed method, thus guaranteeing enough terminal guidance time. The effectiveness of the proposed method is verified by the mooring test-fly experiments.

Focusing on the problem of the millimeter wave (MMW) seeker performance evaluation, which plays an important role for the terminal control algorithm design, an evaluation method is proposed based on the real-time kinematic (RTK) for moving targets. Firstly, time synchronization is realized for different global position system (GPS) carrier platforms taking a controller as the reference. And then, the key parameters associated with the guidance control are calculated on the basis of the GPS measurements. Finally, parameter comparisons are implemented by using the calculated values and the seeker's outputs. The effectiveness of the proposed MMW seeker evaluation method is verified by the mooring test-fly experiments.

For adaptive region-based despeckling methods, the shape of the adaptive region is of large influence on the estimation of the true value. In this paper, given the effectiveness of patch-based index on measuring the relativity of samples, an adaptive region is formed at each pixel via the patch-based index where the local geometric correlation (such as the anisotropy and the compactness) among the pixels in the patch is explored as an Adaptive Kernel(AK) function. Then, using all pixels contained in the adaptive region, Maximum likelihood rule is adopted to estimate the true value of the concerned pixel. From the experimental results on real and synthetic SAR images, by using the AK function embedded patchy index to determine the adaptive region, not only the speckle is largely reduced, but also the resolution of the details is well preserved by our method.

A two-stage sparse structure representation algorithm which can preserve the manifold structure of the data is proposed for synthetic aperture radar target configuration recognition in this paper. Manifold structure of the data is preserved by two stages. In the training stage, taking advantage of both the sparse representation (SR) and manifold learning, local structure of the data is preserved in the reconstruction space, where SR-based recognition is realized. In the testing stage, two structure preserving factors based on the testing samples are embedded into the SR model to enhance structure preserving performance. The first one is constructed to preserve the local structure of the testing samples, which can guarantee the samples that are close to each other in the original space will also be close to each other in the sparse space. And the second one is established to preserve the distant structure of the testing samples, which can ensure the samples that are far from each other in the original space will also be far from each other in the sparse space. Manifold structure of the data is well captured and preserved by two stages. Experimental results on the moving and stationary target acquisition and recognition database demonstrate the effectiveness of the proposed algorithm.

Locality preserving projections (LPP) can preserve the local structure of the datasets effectively. However, it is not capable of separating the samples that are close to each other in the high-dimensional space but belong to different classes. Focusing on the problem, a class-dependent locality preserving projections (CDLPP) algorithm is proposed in this paper. The class information is embedded into the LPP model, and the similarity matrix and the difference matrix are constructed according to the class information. The similarity matrix is utilized to preserve the local structure of the samples belong to the same class, whereas the difference matrix is utilized to separate the samples that are close to each other in the high-dimensional space but belong to different classes. Experiments are conducted using the moving and stationary target acquisition and recognition (MSTAR) database, the results verify the effectiveness of the proposed algorithm.

The performance of the seeker highly influences the design of the control algorithms and the attack precision of the missile. Before the missile with seeker mounted on is launched, the performance of the seeker needs to be accurately evaluated, especially for the expensive ones. Focusing on the problem, a millimeter wave seeker evaluation method is proposed based on the differential global positioning system (DGPS) principle. Firstly, the parameters of the line-of-light (LOS) rates and the missile to target distance are calculated with the data obtained by the DGPS. Then, the results are compared to the ones that are outputted by the seeker itself. The effectiveness of the proposed algorithm is verified on the real seeker data, comparisons with the inertial navigation system (INS) further demonstrate the advantage of the proposed method.