The algorithm of synthetic aperture radar automatic target recognition (SAR-ATR) is generally composed the extraction a set features that transform raw input into representation, followed by trainable classifier. feature extractor often hand designed with domain knowledge and can significantly impact classification accuracy. By automatically learning hierarchies from massive training data, deep convolutional networks (ConvNets) recently have obtained state-of-the-art results in many computer...

Following the great success of deep convolutional neural networks (CNNs) in computer vision, this paper proposes a complex-valued CNN (CV-CNN) specifically for synthetic aperture radar (SAR) image interpretation. It utilizes both amplitude and phase information complex SAR imagery. All elements including input-output layer, convolution activation function, pooling layer are extended to domain. Moreover, backpropagation algorithm based on stochastic gradient descent is derived CV-CNN...

Deep convolutional neural networks have achieved great success in computer vision and many other areas. They automatically extract translational-invariant spatial features integrate with network-based classifier. This letter investigates the suitability potential of deep network supervised classification polarimetric synthetic aperture radar (POLSAR) images. The multilooked POLSAR data format coherency or covariance matrix is first converted into a normalized 6-D real feature vector....

The bidirectional analytic ray tracing (BART) method is developed to rapidly calculate composite scattering from three-dimensional (3D) electrically large complex targets above a randomly rough surface. Ray carried out both along the incident (forward) direction and converse of (backward) recording different orders illumination on each facet or edge target Once pair forward backward rays meet facet/edge, term constructed using diffused scattering/diffraction this facet/edge all reflections...

Synthetic aperture radar (SAR) and inverse synthetic (ISAR) imaging technology are powerful tools to acquire high-resolution image, which is an important basis for further automatic target recognition (ATR). For ISAR, if the frequency high enough, signals that received usually have strong sparsity when data convert Fourier domain, they can be downsampled restored by compressed sensing (CS). As regularization method in CS theory, while L₁ popular, Lq(0 < q 1) proved a sparse framework could...

Reconstruction of 3-D objects from multiaspect high- resolution synthetic aperture radar (SAR) images is great importance for SAR technology applications. In this paper, simple building are modeled as cuboids, and an approach automatic reconstruction in meter developed. The edge detector constant false alarm rate a Hough transform technique parallel line segment pairs first employed to extract the parallelogram-like image walls images. A set probability density functions presented describe...

In this paper, a new All-Convolutional Networks (A-ConvNets) is proposed and applied to Moving Stationary Target Acquisition Recognition (MSTAR) data. Conventional deep learning algorithms, especially the convolutional networks (ConvNets) have achieved many success state-of-art results. However, directly applying ConvNets SAR data will yield severe overfitting because of limited availability. The A-ConvNets can significantly reduce number free parameters degree overfitting. Average accuracy...

Spaceborne high-resolution synthetic aperture radar (SAR) is a potential powerful tool for rainfall pattern and intensity observations over the sea surface. However, many interesting rain-related phenomena revealed by SAR images are still not fully understood due to poor theoretical modeling of rain–wind–wave interactions. This paper attempts develop physics-based radiative transfer model capture scattering behavior rough Raindrops modeled as Rayleigh nonspherical particles, whereas...

With the rapid development of spaceborne synthetic aperture radar (SAR) technology and acquisition a large volume SAR images, image interpretation has become an urgent difficult research topic. statistical modeling is one theoretical foundations for interpretation. It great value in-depth analysis images. This article reviews major developments since its beginning, including more than 20 distributions eight models, gives their derivations expressions, which can be used as basic reference...

Deep learning-based synthetic aperture radar (SAR) target recognition often suffers from sparsely distributed training samples and rapid angular variations due to scattering scintillation. Thus, data-driven SAR is considered a typical few-shot learning (FSL) task. This article first reviews the key issues of FSL provides definition A novel adversarial autoencoder (AAE) then proposed as an representation generation network. It consists generator network that decodes knowledge images...

The tapered wave with the tapering parameter g has been applied to numerical simulation of electromagnetic scattering from randomly rough surface illuminated finite length L. This communication presents how select parameters and L, shows quantitative relationship incident angle. g, as a function angle, should be large enough make equation in range allowable error. larger than several correlation lengths, suppress punctuated power at edges limited efficient computation. Combining these...

A numerical finite-difference time-domain approach for electromagnetic scattering from an object over a randomly rough surface in three-dimensional (3-D) model is developed. Rough truncated simulation by using the periodic extension, and one period of with dependence correlation length used computation. Generation incident wave upon presented. algorithm single on or above are Polarized bistatic obtained based distribution near zone fields. Comparison conventional two-dimensional also discussed.

A hybrid algorithm, combining analytic Kirchhoff approximation (KA) and numerical method of moments (MoMs), is developed to solve the coupling electric-field integral equations (EFIEs) scattering from a perfect electric conducting (PEC) object above randomly rough PEC surface under TE-polarized plane-wave incidence. The MoM with conjugate gradient approach used first EFIE object. fields on are analytically expressed using KA method, large memory computations for those greatly reduced. An...

Simulation of inverse synthetic aperture radar (ISAR) imaging a space target and reconstruction under sparse sampling via compressed sensing (CS) are developed. The numerical bidirectional analytic ray tracing (BART) method is employed to compute the polarized scattering from an electrically large target. With multiorbit multistation modes, 2-D 3-D ISAR images acquired, leading information retrieval target, such as shape, structure, attitude, etc. CS introduced into sampling. As example,...

Numerous radar polarimetry theories and polarimetric synthetic aperture (PolSAR) processing methods have been developed. However, the vast majority of SAR images are not fully (full-pol). This paper proposes "radar image colorization"to reconstruct a full-pol from nonfull-pol image, so that existing PolSAR methods, such as model-based decomposition unsupervised classification, can be directly applied to reconstructed images. It train specially designed deep neural network convert single...

To monitor a space target, 3-D reconstruction from multiview sequence of the inverse synthetic aperture radar (ISAR) imaging is developed. Scattering complex electric-large e.g., ENVISAT satellite model, numerically calculated, and 2-D ISAR can be simulated. Under sparse sampling via compressed sensing, Kanade-Lucas–Tomasi feature tracker applied to extraction target points. Then, using orthographic factorization method, those points produced. A simple hexagonal frustum first tested for...

A reconfigurable 1 bit millimeter-wave array antenna for mobile terminals is investigated. The consists of four microstrip patch elements. Each element fed by a slot on the ground. with phase values <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$-\pi $ </tex-math></inline-formula> and 0 achieved selecting feeding direction line. After analyzing performance four-element array, fixed applied to mitigate...