Weidong Hu

Sparse Representation Based Autofocusing Technique for ISAR Images

From the perspective of sparse signal representation, an autofocusing method in inverse synthetic aperture radar imaging is proposed. Different from the idea of taking the entropy or contrast as the optimization objective in the presently existing algorithms, this method exploits the intrinsic sparsity distribution of scattering centers to compensate the indeterminacy of the measurement system, and a universal regularization model is constructed to simultaneously balance the measurement errors and the sparsity constraint. Accordingly, an effective iterative algorithm on the basis of solving a matrix equation and a trigonometric equation is proposed to estimate the phase errors, which makes the conventional minimum entropy method (MEM) a special case of the proposed method. Specifically, with the sparsity measure being selected as the logarithm function, an analytic representation is derived for the solution of the matrix equation, and the convergence and computational complexity of the proposed method is also discussed. Experimental results show that the proposed method outperforms the present data-driven algorithms in terms of efficiency and robustness, such as MEM, phase gradient autofocusing algorithm, and maximum contrast method.

Numerical Simulation of Vector Wave Scattering From the Target and Rough Surface Composite Model With 3-D Multilevel UV Method

Numerical simulation of vector wave scattering from three-dimensional (3-D) target and rough surface composite model is investigated with a 3-D multilevel UV method. Due to the adoption of RWG basis functions for accurate modeling of vector current, the oscillation of the interaction matrix elements brings difficulty to directly apply the UV decomposition method. Based on the reordering of the interaction strength and the sampling according to the characteristics distribution of the interaction, an EM-interaction-based sampling algorithm is developed for the accurate reconstruction of the far interaction submatrix with UV decomposition method. Combined with multilevel division of the total composite model, the 3-D multilevel UV method incorporating the new sampling algorithm is developed for vector wave scattering from 3-D complex target above or on a random rough surface. The 3-D multilevel UV method yields a complexity of O(N log N) for the setup time of the impedance matrix, the solve time of the matrix iterative solution and also for the memory requirements. The accuracy and the efficiency of the 3-D multilevel UV method is compared and validated with the full MOM method and the ACA method in the tested cases. Finally, the applications of a target above or on the rough surface, for example, a ship on the sea surface, have been accomplished and analyzed.

Automatic Incorporation of Surface Wave Poles in Discrete Complex Image Method

Discrete complex image method is introduced to get a closed-form dyadic Green’s function by a sum of spherical waves. However, the simulation result by the traditional discrete complex image method is only valid in near-field for several wavelengths. In this paper, we analyze the form of spectral domain dyadic Green’s function in the whole kρ plane and the variety of valid range of simulation results by different sampling paths in two-level discrete complex image method. Consequently, for dyadic Green’s function, surface wave pole contribution both in spectral domain and spatial domain is clarified. We introduce the automatic incorporation of surface wave poles in discrete complex image method without extracting surface wave poles. The contribution of surface wave poles in spectral domain and spatial domain dyadic Green’s function is further confirmed in the new

The BCGS-FFT Method Combined With an Improved Discrete Complex Image Method for EM Scattering From Electrically Large Objects in Multilayered Media

This paper presents an efficient algorithm combining the stabilized biconjugate gradient fast Fourier transform (BCGS-FFT) method with an improved discrete complex image method (DCIM) for electromagnetic scattering from electrically large objects in both lossless and lossy multilayered media. The required spatial Greens functions obtained by the improved DCIM are accurate both in the near- and far-field regions without any quasi-static and surface-wave extraction. Then, the scattering by buried objects is considered using the BCGS-FFT method combined with the improved DCIM. Numerical results show the improved DCIM can save tremendous CPU time in scattering involving buried objects.

Range Profile Analysis of the 2-D Target Above a Rough Surface Based on the Electromagnetic Numerical Simulation

In this paper, 1-D range profiles of a 2-D perfect electric conductor (PEC) target above a 2-D PEC rough surface are investigated with numerical solutions for the electric field integral equations (EFIEs) of the combined scattering model. The backscattering is computed accurately by the reradiation of the induced surface currents on the target and the rough surface. Using a stepped frequency waveform (SFW), 1D high-resolution range profiles (HRRPs) of the target above a rough surface are obtained by performing the inverse discrete Fourier transform (IDFT) on the wideband backscattered field. Plots of range profiles show that the multiple interactions between the target and the bottom surface lead to a series of equivalent range profiles, especially when the bottom surface is smooth. Range locations of the equivalent range profiles are in good agreement with the results expected from the ray theory. The range profiles could be understood and analyzed based on the knowledge of the scattering mechanisms. Thus, the connection between the numerically simulated range profiles and the scattering mechanisms is established.

Geodesic Computation on NURBS Surfaces for UTD Analysis

One of the great challenges in calculating the uniform geometrical theory of diffraction (UTD) surface diffracted fields on NURBS surfaces is due to the difficulty in determining the geodesic paths along which the creeping waves propagate. On one single parametric surface patch, the geodesic computation needs to be performed by solving the geodesic equations numerically. Furthermore, realistic objects are generally modeled as the union of several connected NURBS patches. Due to the discontinuity of parameter between patches, it is more complicated to compute geodesic paths on several connected patches than on one single patch. Therefore, this letter develops an adjustable modeling scheme that can adjust the parameterizations of several connected patches to support the geodesic computation throughout these patches. Based on the geodesic computation, the UTD diffractions by NURBS models can be analyzed.

Ray-tracing method for creeping waves on arbitrarily shaped nonuniform rational B-splines surfaces

An accurate creeping ray-tracing algorithm is presented in this paper to determine the tracks of creeping waves (or creeping rays) on arbitrarily shaped free-form parametric surfaces [nonuniform rational B-splines (NURBS) surfaces]. The main challenge in calculating the surface diffracted fields on NURBS surfaces is due to the difficulty in determining the geodesic paths along which the creeping rays propagate. On one single parametric surface patch, the geodesic paths need to be computed by solving the geodesic equations numerically. Furthermore, realistic objects are generally modeled as the union of several connected NURBS patches. Due to the discontinuity of the parameter between the patches, it is more complicated to compute geodesic paths on several connected patches than on one single patch. Thus, a creeping ray-tracing algorithm is presented in this paper to compute the geodesic paths of creeping rays on the complex objects that are modeled as the combination of several NURBS surface patches. In the algorithm, the creeping ray tracing on each surface patch is performed by solving the geodesic equations with a Runge-Kutta method. When the creeping ray propagates from one patch to another, a transition method is developed to handle the transition of the creeping ray tracing across the border between the patches. This creeping ray-tracing algorithm can meet practical requirements because it can be applied to the objects with complex shapes. The algorithm can also extend the applicability of NURBS for electromagnetic and optical applications. The validity and usefulness of the algorithm can be verified from the numerical results.

Investigation of range profiles from buried 3-D object based on the EM simulation

The 1-D range profiles are suitable features for target identification and target discrimination because they provide discriminative information on the geometry of the target. To resolve features of the buried target, the contribution from individual scattering centers of the buried target in the range profiles need to be identified. Thus, the study of complex scattering mechanisms from which the range profiles are produced is of great importance. In order to clearly establish the relationship between the range profile characteristics and the complicated electromagnetic (EM) scattering mechanisms, such as reflections and diffractions, a buried cuboid possessing straight edges is chosen as the buried target in this paper. By performing an inverse discrete Fourier transform (IDFT) on the wideband backscattered field data computed with an accurate and fast EM method, the 1-D range profiles of the buried cuboid is successfully simulated. The simulated range profiles provide information about the position and scattering strength of the cuboids scattering centers along the range direction. Meanwhile, a predicted distribution of the scattering centers is quantitatively calculated for the buried cuboid based on the ray path computation. Good agreement has been found between simulated and predicted locations of the range profiles. Validation for amplitudes of the range profiles is further provided in the research. Both the peak amplitudes and locations of the range profiles could be understood and analyzed based on the knowledge of the scattering mechanisms. The formation of the 1-D range profiles has been revealed clearly from the full analysis of the scattering mechanisms and contributions. The problem has been solved for both near and far field regions. Finally, the buried depth and the characteristic size of the object are reasonably deduced from the simulated range profiles.

Diversity-Promoting Deep Structural Metric Learning for Remote Sensing Scene Classification

Deep models with multiple layers have demonstrated their potential in learning abstract and invariant features for better representation and classification of remote sensing images. Moreover, metric learning (ML) is usually introduced into the deep models to further increase the discrimination of deep representations. However, the usual deep ML methods treat the training samples in each training batch in the stochastic gradient descent-based learning procedure independently, and thus, they neglect the important contextual (structural) information in the training samples. In this paper, we first introduce deep structural ML (DSML) into the literature of remote sensing scene classification and specifically capture and use the structural information during the training on the remote sensing images. Further analysis demonstrates that DSML usually makes many learned metric parameters similar. This similarity leads to obvious model redundancy and thus decreases the representational ability of the model. To address this problem, this paper proposes a new diversity-promoting DSML (D-DSML) method by regularizing the learning procedure by a diversity-promoting prior over the parameter factors. The proposed D-DSML encourages the parameter factors to be uncorrelated, such that each factor can model unique information, and thus, the model’s description ability and classification performance would be significantly improved. Experiments over six real-world remote sensing scene data sets demonstrate that the proposed method obtains much better results than those obtained by the original deep models and has comparable or even better performances when compared with state-of-the-art methods.

Investigation of Range Profiles from a Simplified Ship on Rough Sea Surface and Its Multipath Imaging Mechanisms

The range profiles of a two-dimension (2 D) perfect electric conductor (PEC) ship on a wind-driven rough sea surface are derived by performing an inverse discrete Fourier transform (IDFT) on the wide band backscattered field. The rough sea surface is assuming to be a PEC surface. The back scattered field is computed based on EM numerical simulation when the frequencies are sampled between 100 MHz and 700 MHz. Considering the strong coupling interactions between the ship and sea, the complicated multipath effect to the range profile characteristics is fully analyzed based on the multipath imaging mechanisms. The coupling mechanisms could be explained by means of ray theory prediction and numerical extraction of the coupling currents. The comparison of the range profile locations between ray theory prediction and surface current simulation is implemented and analyzed in this paper. Finally, the influence of different sea states on the radar target signatures has been examined and discussed.