Inverse Synthetic Aperture Radar Image Reconstruction Research Articles

Joint random stepped frequency ISAR imaging and autofocusing based on 2D alternating direction method of multipliers

Random stepped frequency (RSF) inverse synthetic aperture radar (ISAR) has attracted more attention due to its outstanding electronic counter-countermeasure (ECCM) performance by frequency agility. However, the two-dimensional (2D) sparse sampling echo, i.e. sparse sub-pulses and sparse aperture, increases the difficulty of 2D high-resolution ISAR imaging and autofocusing. Aiming to solve this problem, a novel 2D alternating direction method of multipliers (2D-ADMM) based imaging and autofocusing framework is proposed to achieve 2D high-resolution for 2D sparse RSF ISAR. In order to obtain the optimal value for phase correction, the phase error corresponding to each sub-pulse is estimated by solving an unconstrained optimization problem. To jointly achieve autofocusing during the process of ISAR image reconstruction, the phase error estimation is merged into the image reconstruction framework, and then two different strategies under the 2D ADMM framework are derived to iteratively solve this compound optimization problem in matrix form. Experiments based on both simulated and measured data validate the effectiveness of the proposed 2D joint methods, meanwhile, maintaining an acceptable imaging efficiency.

ISAR imaging enhancement: exploiting deep convolutional neural network for signal reconstruction

ABSTRACT Since the radar imaging can serve as an ill-posed inverse problem, it is an appealing topic to introduce deep learning instead of regularized iterative algorithm to address the presence of uncertainties in the inverse synthetic aperture radar (ISAR) forward model. Usually, the obtained image quality is largely limited by the sparse representation and denoising performance of the target scene. This paper presents a deep learning technology for radar echo signal recovery with enhanced ISAR imaging capability. If the convolutional neural network (CNN) is regarded as an imaging processor, its input and output are echo data and formed image, respectively. The trained CNN contains a variety of feature extractors, which can automatically extract and abstract high-level image feature representation. The main difficulty of CNN applications in the radar imaging field, in fact, is that a large amount of appropriate training data cannot be obtained as easy as in other fields. A novel deep CNN (DCNN)-based method, U-net-based imaging, is proposed for ISAR image reconstruction using less training samples compared to existing CNN imaging networks. Some improvements are presented to make it adapt to imaging tasks. Experimental simulations are carried out to prove the remarkable performance of the proposed method. In the presence of random noise and echo undersampling, the trained U-net achieves fast and accurate reconstruction of high-quality ISAR images, respectively.

Three-dimensional ISAR image reconstruction technique based on radar network

ABSTRACT Recently, there are few reports on three-dimensional (3D) imaging method based on radar network. Aiming at the 3D imaging problem in radar network, a reconstruction method for 3D inverse synthetic aperture radar (ISAR) image based on radar network is proposed in this paper. The 2D ISAR image is considered as a geometric projection of a target on an imaging projection plane; thus, aerial points might be scatterers of targets if their geometric projection on an imaging projection plane is similar to the 2D ISAR image of the target. The proposed method makes use of the similarity between the ISAR Range-Doppler (RD) image and the target’s geometric projection on the imaging projection plane to construct the 3D image reconstruction model. The reconstructed target 3D ISAR image is obtained by estimating the scattering centre position in 3D Cartesian space. In this paper, the projection formulas designed to calculate the geometric projection of the aerial points on the imaging projection plane are derived to first form the projection image. Depending on the similarity between the 2D ISAR image and the target’s projection, the 3D ISAR image reconstruction model based on the radar network is constructed. Further, a corresponding algorithm for solving the 3D ISAR image reconstruction model is proposed. Finally, the radar distribution influence on reconstruction precision is analysed, and some experimental results are given to illustrate the effectiveness of the proposed method.

Compressive sensing approach for high‐resolution ISAR image reconstruction and autofocus

In this study, a novel autofocus method is presented to achieve high-resolution inverse synthetic aperture radar (ISAR) image reconstruction with limited measurements in the compressive sensing (CS) framework. In order to solve the CS reconstruction problem of ISAR images with sparsity, the CS coupled with Tikhonov-regularisation-based algorithm is devised as an effective focusing technique. The proposed method aims at compensating the phase errors induced by target motions, and can provide globally well-focused ISAR images. Simulation and experimental results verify the effectiveness of the proposed scheme.

An Efficient Method on ISAR Image Reconstruction via Norm Regularization

Inverse synthetic aperture radar (ISAR) imaging technology is a powerful tool to distinguish between targets with different types. The high-resolution radar image acquisition is an important basis for further automatic target recognition. In the high-frequency radar band, since the spatially collected received signals have strong sparsity in the image domain (Fourier domain), they can be downsampled and restored by corresponding norm regularization framework. When the target is electrically large, its contributions are equal to several point scatterers [also called scattering centers (SCs)]. So that the imaging process can be regarded as the summation of the point scatterers impulse response, the convolution process, and deconvolution method, the reverse process is adopted to extract the SCs from the image, where the number of SCs reflects the sparseness of ISAR image. As for the reconstruction method, while L <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1</sub> regularization is widely used, the L <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">q</sub> (0 <; q <; 1) regularization has proved to be a more sparse regularization framework and achieves better performance in reconstructing a signal with a lower down-sampling rate. This article implements L <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1</sub> regularization into the ISAR image reconstruction 2 under down-sampling. The ISAR image is generated by patch integration formula from the physical optics method. The sparsity of the ISAR image is estimated through the appropriate deconvolution process CLEAN. Given the sparsity estimated, and compared with the L <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1</sub> regularization, the experiment of two complex geometry targets validates the reconstruction precision and convergence and lossless recovery probability of L <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">1/2</sub> regularization.

ISAR Image Reconstruction with Heavily Corrupted Data Based on Normal Inverse Gaussian Model

Inverse synthetic aperture radar (ISAR) is a potent radar system which generates two-dimensional signal on the range-Doppler domain by using target’s motion. ISAR imaging of targets is an important tool for automatic target recognition and classification in the defense and aerospace industry. In this paper, we focus upon the problem of ISAR imaging at low signal to noise ratio (SNR). Nonsubsampled directional filter bank (NSDFB) is a very useful tool in studying the directional features in two-dimensional signals. This paper offers a novel ISAR imaging approach by using NSDFB coefficients modeling. Bayesian maximum a posteriori is used where normal inverse Gaussian model is presumed for estimating ISAR image at low SNR. We applied NSDFB transform to ISAR image and implemented procedure to describe the characteristics of the algorithm. Both simulated and real ISAR data have been tested. The proposed technique keeps a balance between feature preservation and noise suppression. Finally, experimental results show that the offered technique outperforms other in terms of visual assessment and image evaluation parameters.

ISAR imaging and cross‐range scaling of high‐speed manoeuvring target with complex motion via compressive sensing

For targets with extreme manoeuvres, inverse synthetic aperture radar (ISAR) imaging suffers from translational motion (TM), which is modelled as a one-dimensional (1D) phase error, and non-uniform rotational motion (RM), which is a multidimensional (MD) phase error that causes severe blurring in ISAR images. Full-aperture data collection is often unachievable because of interference with other radar activities, resulting in sparse-aperture (SA) data. In this study, the authors present a new framework for SA-ISAR imaging and cross-range scaling for manoeuvring targets based on compressive sensing. Instead of solving conventional optimisation problems constrained by a sparsity of signals, the proposed method utilises the sensing-matrix estimation technique for ISAR image reconstruction using parametric signal-model reconstruction. To do this, it looks for basis functions that best represent the behaviour of a sensing-dictionary matrix comprising the observed SA data. The sensing-matrix reconstruction is based on a modified orthogonal matching pursuit-type basis function-searching scheme. Finally, they generate a well-focused and scaled ISAR image from the recovered complete ISAR signal using the conventional Fourier transform after the removal of signals corresponding to 1D TM and MD RM phase errors. They utilise both simulated and real measured datasets to confirm the effectiveness of proposed method.

Enhanced ISAR Imaging and Motion Estimation With Parametric and Dynamic Sparse Bayesian Learning

This paper is focused on high-resolution inverse synthetic aperture radar (ISAR) imaging and motion estimation of maneuvering targets from compressively sampled echo data. Herein, a local structural sparse Bayesian learning (LS-SBL) algorithm is proposed by exploiting the joint sparsity pattern of adjacent scatterers. A structured prior by modeling the neighboring correlation or dependence is utilized to encode the joint sparsity pattern. Meanwhile, a parametric dictionary with unknown rotational parameters is constructed to represent the target maneuverability. The solution to the LS-SBL algorithm is decomposed into iterations between sparse imaging and dictionary learning. In sparse imaging, an expectation-maximization method is employed for ISAR image formation and hyperparameter estimation by using a predesigned dictionary. In dictionary learning, an efficient approach of rotational parameter estimation is presented to dynamically update the parametric dictionary. Due to the exploitation of joint sparsity pattern, enhanced performance of ISAR image reconstruction can be achieved by effectively preserving the target structure. In addition, the cross-range scaled ISAR image is obtainable by extracting the target geometry, which benefits from the rotational motion estimation. Finally, experiments on simulated and measured data demonstrate the effectiveness of the proposed algorithm.

Bistatic three-dimensional interferometric ISAR image reconstruction

A three-dimensional (3D) bistatic inverse synthetic aperture radar (ISAR) imaging method is proposed in this paper. The proposed method makes use of interferometry and technically speaking, produces a 3D target reconstruction by estimating a scattering center position in 3D Cartesian space. The proposed method makes use of a combined ISAR/interferometry technique that also allows the ISAR image plane to orientation to be estimated. Cross- or L-shaped antenna configurations are discussed, and the effects of the baseline length along the horizontal and vertical direction on the scatterer’s position estimation are analyzed in detail. Finally, numerical simulations are used to evaluate the proposed method’s performance.

Improved compressed sensing for high-resolution ISAR image reconstruction

For inverse synthetic aperture radar (ISAR), an ISAR signal in the cross-range direction has the characteristic of sparsity in the azimuth frequency domain. Due to this property, a Fourier basis is adopted as a kind of sparse basis, and high cross-range resolution imaging is achieved by using the compressed sensing (CS) method. However, the Fourier expanding for signal with finite length will result in energy leaking and spectrum widening. As a result, the Fourier basis cannot obtain the optimum sparse representation for signals of unknown frequencies in most cases. In this paper, we present an improved Fourier basis for sparse representation of the ISAR signal, which is constructed by frequency shift and weighting of the Fourier basis and available to obtain the robust recovery performance via CS. Simulation results show that the improved CS method outperforms conventional CS method that uses the Fourier basis.

Two-dimensional ISAR model and image reconstruction with stepped frequency-modulated signal

A two-dimensional (2-D) inverse synthetic aperture radar (ISAR) return signal model that employs stepped frequency (SF) modulation is developed. The geometry of the examined ISAR scenario is described by analytical geometrical equations. The target to be imaged is represented by a rectangular grid of point scatterers, moving along a rectilinear trajectory at constant speed, without any rotational motion. Thus, the inverse synthetic aperture results from the translational motion of the target for a short period of time. The process of ISAR signal modelling through coherent summation of the SF-modulated signals reflected from different point scatterers of the target is thoroughly described. Moreover, an efficient ISAR image reconstruction approach, including cross-correlation-based range compression and fast-Fourier-transform-based azimuth compression, is presented through analytical mathematical expressions. Numerical simulations are carried out for various SF ISAR scenarios and high-resolution ISAR images are obtained by applying the proposed ISAR image reconstruction approach. Simulation results (ISAR images and corresponding entropy values) indicate the validity of the proposed 2-D SF ISAR return signal model and the efficiency of the proposed imaging algorithms. Finally, a numerical simulation result is illustrated, which shows the comparison of the performance of the proposed ISAR image reconstruction algorithms based on SF and linear frequency modulation waveforms. It is shown that the two waveforms attain almost the same ISAR image resolution.

Time windowing for highly focused ISAR image reconstruction

In several applications long recorded live inverse synthetic aperture radar (ISAR) data are used to obtain one or more ISAR images. In order to reconstruct a well-focused ISAR image, a suitable selection of the echoes to be coherently processed must be provided. Such a selection can be made by defining a time window. We propose a technique for the automatic selection of the position and length of the time window that provides the ISAR image with the highest focus. The technique, namely the maximum contrast based automatic time window selection (MC-ATWS), is based on the definition of image contrast (IC). Due to the fact that the IC is a measure of the image focus, the time window is selected by maximizing the IC. The technique effectiveness is tested by using simulated and real data.

3d iSar Image Reconstruction of a Target With Motion Data Using Adaptive Feature Extraction

In this paper, three-dimensional (3D) inverse synthetic aperture radar (ISAR) image reconstruction with known motion data is studied. In traditional two-dimensional (2D) ISAR imaging, a 2D point scatterer model is adequate to consider the target rotation motion with a fixed rotational axis. However, target motions with a varying rotational axis are sometimes encountered in real situations. Under such cases, the use of a 3D point scatterer model is necessary for 3D ISAR image reconstruction. An adaptive feature extraction algorithm is proposed to reconstruct the 3D image of a target with non-uniformly undersampled radar data over the azimuth and elevation aperture. Simulation results based on actual motion data of air targets demonstrate the effectiveness of the algorithm.

Iterative MMSE method and recurrent Kalman procedure for ISAR image reconstruction

This work presents a novel approximate iterative and recurrent approach for image reconstruction from inverse synthetic aperture radar (ISAR) data. Mathematical models of the quadrature components of the ISAR signal, reflected by an object with a complex geometry, are devised. Approximation matrix functions are used to describe deterministic signals reflected by point scatterers located at nodes of the uniform grid (model) during inverse aperture synthesis. Minimum mean square error (MMSE) equations and Kalman equations are derived. To prove the validity and correctness of the developed iterative MMSE method and recurrent Kalman procedure, numerical experiments were performed. The computational results demonstrate high resolution images, unambiguous and convergent estimates of the point scatterers' intensities of a target from simulated ISAR data.

Multipath effects on ISAR image reconstruction

The effects of specular multipath on the reconstruction of high resolution target image provided by an inverse synthetic aperture radar (ISAR) are investigated. First, a model of the received signal that takes into account the reflections produced by the surface of the Earth is defined, Then, the impulse response of the imaging system used to reconstruct the reflectivity function of the target is evaluated. Finally, the theoretical analysis is validated by some simulation results relative to different motion conditions.