Over past few years, the development of deep learning has greatly facilitated the SAR target recognition tasks. However, most deep learning-based SAR aircraft recognition methods suffer from a drawback of heavy reliance on large-scale labeled training data, making it difficult to apply them to actual remote sensing tasks. In term of these issues, we propose an improved prototypical network based on two stage fine-tuning for few shot SAR aircraft recognition. In this method, firstly, the deep learning image encoder is trained on large scale open source datasets supervised. This gives the encoder the ability of recognizing basic geometry. Secondly, based on the contrastive learning, the encoder is trained on SAR aircraft data in a self-supervised manner. This helps the encoder learn the ability to encode the semantics of scatterers in SAR image. Finally, the encoder is further fine-tuned on the specific SAR aircraft few shot recognition dataset in the way of prototype network. The comparative experiments shows that, the proposed method outperforms other state-of-the-art methods on public SAR aircraft dataset. Especially in the 1-shot case, the proposed method achieves a recognition accuracy of 81.34%. Simulation results show that the proposed method has the potential to be applied to real-world scenarios.

In frequency diverse array multiple-input–multiple-output (FDA-MIMO) radar for target detection, besides range migration and Doppler migration, serious Doppler spread in the slow-time domain also will be a problem because the frequency offset between transmitting arrays is coupled with the target velocity, which results in the signal energy of each receiving channel not being coherently accumulated to reduce the detection performance. To address this issue, this letter proposes a method for FDA-MIMO radar moving target detection using planar array. The moving target returned signal of FDA-MIMO radar using planar array is established. After multi-channel matched filtering, resampling in slow-time domain and phase compensation functions of the acceleration and velocity ambiguity factor are applied to correct the range migration and Doppler migration, and compensate for the Doppler spread caused by frequency offset. In doing so, the target’s energy is coherently accumulated. Simulation results verify the effectiveness of the proposed algorithm. Under the same parameter conditions, in order to achieve a 90% detection probability, the proposed algorithm allows the signal-to-noise ratio (SNR) of input echo signal reduction of at least 3dB.

Image despeckling is important to synthetic aperture radar (SAR) image restoration and related downstream tasks. Due to the fact that existing SAR image denoising algorithms are difficult to simultaneously achieve high performance, efficiency and interpretability, in this paper, we propose a new image denoising method, namely, Trans-NLM, which incorporates the Transformer architecture into traditional nonlocal means filtering (NLM) based denoising algorithm. In doing so, the SAR image despeckling performance is enhanced, while the algorithm interpretability is retained. First, each pixel and its surrounding pixels are simultaneously mapped to a high-dimensional space as a neighborhood matrix through a shallow fully connected layer. Then, positional encodings are added to the neighborhood matrix, which is further mapped to the internal vectors Key, Query and Value. The final vector representation of each pixel is also calculated according to the multi-head attention mechanism. Finally, the representation vector is passed through a shallow fully connected layer for data dimension reduction to predict the corresponding pixel value. Moreover, layer normalization and residual learning are applied to accelerate the convergence. Experiments on both simulated and real SAR data demonstrate that compared with representative denoising models, e.g., NLM, fastNLM, SAR-CNN, CNN-NLM and MONet, the proposed Trans-NLM exhibits better performance in despeckling enhancement and efficiency simultaneously, along with more explainable inference process and transfer learning capability.

Since the deep learning based SAR despeckling models rely heavily on the labeled training data, and struggle to process noisy images with varying noise distribution, this paper proposes an unsupervised SAR despeckling model based on the diffusion model which consists of a forward and a reverse processes. In the forward process, the noise with Gaussian distribution is gradually added to the clear image in the logarithmic domain until the image is heavily contaminated. Then in the reverse process, the noise of the image is gradually predicted and removed by the U-net like neural network until the image is close to the clear image. Furthermore, this paper proposes a shifting and averaging based algorithm for processing high resolution image in patches separately, which gets rid of the dependence on high video memory GPUs. Experiments results demonstrate that the proposed unsupervised despeckling model can be adopted to despeckle SAR images with varying noise intensities simply by adjusting the external parameter values. Though the model’s training does not depend on any clear SAR images, it has close performance compared with advanced supervised models.

In this paper, a new method for solving target range ambiguity in radar signal processing based on subarray FDA-MIMO is proposed. At the same time, The optimal frequency shift selection of the transmit array is also analyzed. Compared to the FDA-MIMO radar with linear frequency shift, the new method is able to exclude the effect of ambiguous range peaks in FDA detecion and has a better result in supressing noise. The new method also extend the maximum unambiguous range, so it is more advantageous in the detection of multiple targets.

In this letter, a frequency diverse array (FDA) is introduced into synthetic aperture radar ground moving target indication (SAR GMTI) to mitigate both the blind velocity and Doppler ambiguity problems. Due to the$2\pi $periodicity of the echo phases, notches will occur periodically in clutter cancelers at nonzero velocities, leading to the blind velocity problem. In the proposed scheme, the dependence of the measured blind velocity on the transmission frequency is considered, and a new clutter canceler unaffected by the blind velocity problem is constructed via the integration of multiple cancelers with diverse frequencies. Moreover, to resolve the Doppler ambiguity, along-track interferometry (ATI) based double interferometry is proposed to extend the maximum unambiguous radial velocity (RV). Additionally, search-based clustering is adopted to enhance the precision of RV estimation. Finally, a moving target can be brought into focus and correctly relocated using the estimated RV. Numerical results verify the effectiveness of the proposed method.

Ground moving target indication (GMTI) has been extensively applied in modern warfare. In this letter, a novel deceptive jamming technique for countering multichannel synthetic aperture radar (SAR)-GMTI is presented, which is implemented with the frequency diverse array (FDA). Different from the phased array (PA), FDA modulates the transmitting signal with multiple frequencies across its array elements, which, consequently, enables range-dimension false target deception for SAR imaging. Furthermore, to achieve effective deception for the GMTI, micromotion modulation is adopted to simulate dynamic features of the moving targets. Mathematical derivation and simulation results show that the proposed technique can efficiently produce massive false targets in both the range and azimuth dimensions.

In this letter, we consider a target with high-order motion and propose two long-time coherent integration algorithms with low computational cost for target detection via phase compensation and time-reversal transform. Due to the even/odd characteristic, the coupling effect between the range frequency and slow time can be eliminated by constructing the phase compensation function and doing the time-reversal transform. In doing so, the target energy can be accumulated by inverse fast Fourier transform (IFFT) in range frequency and fast Fourier transform (FFT) in slow time. Simulations are given to demonstrate the effectiveness of the proposed algorithms. Moreover, compared with the generalized Radon–Fourier transform (GRFT) and generalized dechirp-keystone transform (GDKT), the proposed algorithms have much lower computational burden.

Frequency diverse array (FDA) radar can form range-angle dependent beampattern, which has good anti-jamming ability. When the direction of the interference fluctuates and the desired signal steering vector mismatches, the traditional null broadening algorithm based on phased array (PA) radar can only widen the null in the angle dimension. A null optimization method with adaptive beamforming for FDA is proposed in this paper. The method is based on a uniform linear array and can form a range-angle two-dimensional adaptive beam. The main lobe of the beam is aimed at the target and the null is aimed at the interference. Then the covariance matrix taper (CMT) algorithm is used to broaden and deepen the null at the position of the interference. Experimental results illuminate the feasibility and effectiveness of the proposed method.

Deception against inverse synthetic aperture radar (ISAR) has been a topic of active research in electronic warfare. However, most of the studies mainly focus on producing one-dimensional deception, which is inadequate for two-dimensional imaging system like ISAR. Thus, in this letter, a joint two-dimensional ISAR deception based on frequency diverse array (FDA) and interrupted sampling is proposed. Compared with conventional phased-array (PA) and single-channel methods, FDA permits more degrees of freedom in producing range deception. Combined with interrupted sampling in the slow time domain, the proposed method is anticipated to produce a group of false targets in both range and cross-range directions. The benefit is easy to control the number and distribution of false targets with high efficiency and less computation burden. Simulation results verify the effectiveness of the proposed method.

In this paper, the sparse stepped frequency synthetic aperture radar (SAR) imaging under basis mismatch is investigated. The traditional CS imaging quality decreases significantly during the target point being not on the grid point. Thus an adaptive basis constructed by the frequency shift within one grid is proposed to reduce the mismatch error and recover the off-grid targets accurately. An efficient algorithm of orthogonal matching pursuit is applied in this paper to solve the sparse regularization problem. Simulation results based on two-dimensional sparse data verify the feasibility of the proposed method.

Since many target features can be acquired via inverse synthetic aperture radar (ISAR), the key of implementing deception jamming on ISAR lies in inducing decoy images imitating the scattering features of real target. In this paper, based on the combination of the frequency diverse array (FDA) and scatter-wave jamming (SWJ), an approach producing multiple two-dimensional decoy images with real target features is presented, the distribution and number of which can be arbitrarily manipulated through designing FDA parameters. Details of the jamming on ISAR is provided along with the simulation results, which illuminates the feasibility and validity of the proposed method.

This paper is devoted to the research of methods and experiments of regularization deconvolution theory on the azimuth super-resolution of forward-looking imaging radar. L1 norm is usually used as a regular term to obtain a stable solution due to its strong resolving power for sparse targets. However, deconvolution is an ill-posed problem, in the process of deconvolution iteration, using the L1 norm as a regular term is sensitive to noise and may causes a large deviation between the solution and the true value due to the influence of noise. This paper proposes the super-resolution imaging method combined truncation singular value decomposition (TSVD) with L1 norm constraint. At lower SNR, this method effectively solves the problem of noise amplification during deconvolution iteration. The effectiveness and advancement of the proposed algorithm are verified by simulation results.

Most of the existing deception jamming methods can only achieve effective jamming on one dimension. Thus, a two-dimensional deception jamming method for SAR imaging based on the frequency diverse array (FDA) and micro-motion modulation is proposed in this paper. The proposed method can produce a number of false targets in both the range and azimuth of SAR image, the location and number of which can be specifically controlled. Based on SAR imaging geometry, principles of the jamming are derived. According to the final image expression, jamming performance with different parameters are demonstrated. Experimental results illuminate the feasibility and effectiveness of the proposed method.