A novel semantic-aided feature generative adversarial network (SFGAN) is proposed in this letter and utilized in zero-shot learning (ZSL) for fast synthetic aperture radar (SAR) automatic target recognition (ATR). By exploiting shared semantic attributes of all classes, the SFGAN is capable of directly generating and recognizing samples with sufficient discriminability and reality from feature dimension, and ultimately achieves accurate and fast recognition of unseen targets. To enhance the quality and stability of feature space, the Wasserstein distance measurement based on the semantic attributes is applied in generative adversarial network (GAN). Meanwhile, in order to improve the discriminability of generated features, a classification sub-net is integrated into the network. Additionally, Gaussian mixture model (GMM) and reconstruction loss are introduced in feature generation to align the unseen features more consistent with distribution of real data. In experiments on a new self-built SAR aircraft dataset, SFGAN achieves 70.14% accuracy on unseen targets and improves the training speed by more than two times, fully demonstrating the advancement and effectiveness of the proposed method.

A multiband fusion (MF) method can estimate a full-band echo (FBE) with a bandwidth larger than the sum of bandwidths of several subbands echoes. To improve estimation speed and estimation accuracy of the FBE, a novel MF method based on a small multiband-measurement matrix and a nonconvex log-sum regularization (LSR) is proposed and experimentally verified. In the method, the multiband-measurement matrix, which is in a subbands-echoes expression, provides a beneficial tool for speeding up the estimation of the FBE due to its small size. The small size is obtained by constructing the matrix only using the parameters of existing subbands and ignoring parameters of gaps between subbands. The LSR is developed in a multiband cost function to improve the estimation accuracy of the FBE, for its corresponding geometric property is similar to a geometric property of an$l_{0}$-norm regularization that can guarantee an unbiased estimate of FBE. In addition, since the imaging scene is sparse and the lack of several pulses does not affect the sparsity, the proposed method is robust to missing pulses. Simulation-data and real-date experimental results proved that the bandwidth of a fused FBE is 1.67 times the sum bandwidth of two subbands, the estimation accuracy and the estimation speed corresponding to the proposed method are 5.56 times and 1.11 times those of the traditional MF methods, respectively, and the proposed MF method is still feasible even when 10% of pulses are randomly missing.

In traditional multiband fusion (MF) methods, the differences in the scattering center (SC) among different sub-bands are ignored, which leads to incoherence between sub-band echoes and results in limited estimation accuracy for full-band echo (FBE). To solve this problem, a novel MF method based on differential processing of the SCs in each sub-band is proposed and experimentally demonstrated. In this method, the SCs in each sub-band are classified using a proposed two-layer contrast strategy into two kinds of SCs, which are fused and super-resolved, respectively, to produce the FBE. For application during the classification, fusion, and super-resolution processes, a general iterative shrinkage thresholding algorithm is developed to ensure both the accuracy and speed of estimation of the SC locations and amplitudes. Total FBE is calculated by superimposing the fused FBE and the super-resolved FBE. In experiments, the bandwidth of each sub-band is improved by an average factor of 3.97 when the signal-to-noise ratio is equal to 10 dB and the frequency gap between two sub-bands is equal to 5.08 GHz. Thus, the high estimation accuracy of the proposed method is verified, as well as its high robustness to noise and incoherence.

In this paper, a novel pulse-agile waveform design based on random frequency modulation (RFM) waveforms is proposed for range sidelobe suppression and range ambiguity mitigation to improve the system performance of radar detection and imaging. The time-frequency relationships of the proposed RFM waveforms are characterized by random Fourier coefficients, which are different from other RFM waveform models. To determine the number and the range of random Fourier coefficients, an RFM waveform design framework based on the theoretical derivation of spectral distribution is presented. The proposed framework has the ability to generate spectral-compact pulse-agile RFM waveforms with high efficiency even under a large waveform number and a large time-bandwidth product. The designed RFM waveforms have random sidelobe, good quasi-orthogonality, high spectral compactness, and constant amplitude. By exploiting the advantage of pulse-agile RFM waveforms, range sidelobes can be reduced by the plural accumulation of the pulse compression results of multiple RFM waveforms. Meanwhile, range ambiguity can be mitigated using different filter banks for the corresponding range intervals. Compared with other low sidelobe quasi-orthogonal waveforms, the designed pulse-agile RFM waveforms using random Fourier coefficients perform better in peak-to-sidelobe level ratio (PSLR) and cross correlation peak (CCP) levels with a large waveform number. The PSLR and the average CCP level for a thousand RFM waveforms are -49.0 dB and -25.8 dB, respectively. Finally, a series of simulations and experiments of microwave imaging that transmit pulse-agile RFM waveforms are conducted, which demonstrate the ability of range sidelobe suppression and range ambiguity mitigation.

The difficulty in designing quasi-orthogonal radar waveforms is to ensure low levels of autocorrelation sidelobe peak (ASP) and cross-correlation peak (CCP) under a large number of waveforms. This letter proposes a waveform model with reciprocating nonlinear frequency modulation (RNLFM) for the quasi-orthogonal waveform design. The RNLFM waveform both enjoys the low sidelobe characteristics of nonlinear frequency modulation (NLFM) and high degrees of freedom from reciprocating frequency modulation. A greedy optimization framework is presented to decompose the waveform design problem into two optimization problems that minimize ASP and CCP levels, respectively. Simulation results show that the ASP and the CCP levels of the optimized RNLFM waveforms are reduced by 5.5~9.5 dB and 0.2~2.4 dB respectively, compared with other quasi-orthogonal waveforms under the same waveform number.

The nonlinear frequency modulation (NLFM) waveform can achieve low sidelobes without loss of signal-to-noise ratio (SNR) by utilizing a nonlinear time-frequency structure. This letter presents a novel two-step progressive optimization framework for the NLFM waveform using Fourier series by exploiting the advantages of random searches and deterministic modifications. The modified GA is used for the preliminary design of the NLFM waveform, where random searches with waveform diversity and consideration of the signal’s characteristic provide a rapid decrease of sidelobes in early iterations. When the descent speed of the sidelobes slows down, deterministic modifications in the gradient-based tuning are followed to further reduce the sidelobes. In the simulation, the peak-to-sidelobe level ratio (PSLR) is reduced by at least 5 dB with an almost equivalent normalized impulse response width (IRW) under the same time-bandwidth product (TBP) compared with other NLFM waveforms. Finally, the optimized NLFM waveform is implemented on hardware, which verifies its performance.

Microwave photonic synthetic aperture radar (MWP-SAR) offers a larger signal bandwidth than conventional SAR, and its theoretical resolution can be improved to centimeter level. To achieve same order of magnitude resolution in azimuth direction, long synthetic aperture is always required, which results in extremely high requirements for the accuracy of imaging procedure. Compared with conventional SAR imaging algorithms, two major problems should be considered: 1) The scattering characteristics of target might vary from the frequency of signal and the angle of incidence, which seriously affects the coherence of received echo. 2) The imaging of MWP-SAR system is more sensitive to motion errors and therefore requires higher accuracy of motion compensation processing. To solve the above issues, a high-precision imaging method for MWP-SAR is proposed. First, based on the attribute scattering center(ASC) model, this paper analyzes the target-scattering characteristics with different frequencies and incident angles. Then, according to the influence of scattering phase on MWP-SAR imaging, an optimal sub-aperture division algorithm is proposed to guarantee the coherence of each sub-aperture data and better imaging results. Furthermore, to compensate for the effects of high-order motion errors, this paper proposes a motion error estimation algorithm based on sub-image registration, where the full aperture high-order motion errors can be divided into multiple linear components, and the flight trajectory of platform can be accurately reconstructed. Finally, a full-aperture time-domain high-precision imaging method is presented, and the effectiveness of the proposed formation is verified by both simulation and actual airborne MWP-SAR data processing.

A novel multiband fusion method based on a modified RELAX algorithm (MRA) is proposed. In the MRA, a maximum difference criterion is applied to the singular values of echoes’ Hankel matrix to improve the accuracy of calculating the number of scattering centers. In addition, a transformation process is put forward to merge the frequency-dependent factor (FDF) term, which is in the geometrical theory of diffraction (GTD) model, into the phase term of the echo to simplify the model. Moreover, a matching relationship for the correspondence between the FDF values and the cost function is established to effectively increase the estimated precision of the FDF. Therefore, based on the MRA, exact estimations of the incoherent errors among subband echoes are achieved, and then a fused full-band echo (FBE) with high estimation accuracy is produced. Using all the fused FBEes, high-resolution, anti-non-Gaussian colored clutter microwave imaging is realized. The effectiveness of the proposed method is validated with the simulated and real-data experimental results in a simulated non-Gaussian colored clutter environment.

With the exploration of synthetic aperture radar (SAR), the requirements for its functionality, precision, and response time are inevitably increasing, in which the technical core is the generation, reception, and processing of wideband signals with low time cost. Owing to the excellent performance of modern photonics, such as the ultra-wide bandwidth (UWB), flat response, low transmission loss, fast analog signal processing, and microwave photonics (MWP) promises to be an appropriate solution to improve the capability of SAR in resolution, coverage, and efficiency. However, on the one hand, due to the high sensitivity of optical fiber, moisture, temperature, or physical vibration can produce an unknown extra propagation delay. On the other hand, the wavelength shift effect should be taken into consideration for the UWB system. Thereby, under very high resolution (VHR) circumstance, two-dimensional (2-D) defocus including migration through resolution cells (MTRC) and azimuth phase error (APE) becomes a challenge for MWP SAR imaging. Unfortunately, existing 2-D autofocus approaches concerning motion errors inherently fail for the commonly underlying assumption that the prominent nonsystematic residual range cell migration (RCM) is global in the azimuth time domain. In this article, we analyze the effects of the above-mentioned negative factors in the image processing of range migration algorithm (RMA) and reveal the structural characteristics of the 2-D phase errors. A novel two-step postprocessing compensation strategy is developed, and experiments on real data acquired by an airborne MWP SAR system demonstrate its effectiveness.