Haihua Feng

Moderately rough surface underground imaging via short-pulse quasi-ray Gaussian beams

An adaptive framework is presented for ultra-wideband ground penetrating radar imaging of shallow-buried low-contrast dielectric objects in the presence of a moderately rough air-soil interface. The proposed approach works with sparse data and relies on recently developed Gabor-based narrow-waisted quasi-ray Gaussian beam algorithms as fast forward scattering predictive models. First, a nonlinear inverse scattering problem is solved to estimate the unknown coarse-scale roughness profile. This sets the stage for adaptive compensation of clutter-induced distortion in the underground imaging problem, which is linearized via Born approximation and subsequently solved via various pixel-based and object-based techniques. Numerical simulations are presented to assess accuracy, robustness and computational efficiency for various calibrated ranges of problem parameters. The proposed approach has potential applications to antipersonnel land mine remediation.

A curve evolution approach for image segmentation using adaptive flows

In this paper, we develop a new active contour model for image segmentation using adaptive flows. This active contour model can be derived from minimizing a limiting form of the Mumford-Shah functional, where the segmented image is assumed to consist of piecewise constant regions. This paper is an extension of an active contour model developed by Chan-Vese. The segmentation method proposed in this paper adaptively estimates mean intensities for each separated region and uses a single curve to capture multiple regions with different intensities. The class of imagery that our new active model can handle is greater than the bimodal images. In particular, our method segments images with an arbitrary number of intensity levels and separated regions while avoiding the complexity of solving a full Mumford-Shah problem. The adaptive flow developed in this paper is easily formulated and solved using level set methods. We illustrate the performance of our segmentation methods on images generated by different modalities.

GPR imaging approaches for buried plastic land mine detection

This paper explores new imaging approaches for detection of buried plastic mines using data from a ground penetrating radar array. The first approach is based on fine-resolution imaging of the perrnittivity contrast in the region of interest. We develop a model relating the collected data to the contrast generated by buried objects, using ray optics to represent the air/soil interface and Borrt weak scattering approximations. We use this model to develop inversion algorithms for image formation, using alternative regularization approaches to overcome modeling error and ill-posedness due to limited sensing geometries. The second approach is based on tomographic curve evolution techniques, which use object boundaries and a set of contrast coefficients to represent the underlying perrnittivity contrast field. The inversion method seeks to extract less information than a full image, concentrating on accurate estimation of object boundaries. The problem of determining the reduced parameters is formulated as a non-linear estimation problem, which is solved iteratively using level set techniques. The algorithms are tested on data generated by nonlinear finite difference time domain electromagnetic simulations, under conditions involving different ground roughness and object geometries.

Underground imaging based on edge-preserving regularization

Develops approaches for imaging weak-contrast buried objects using data from a ground penetrating radar array. An approximate physical model relating the collected data to the underground objects is developed. This model uses ray optics to represent the air/soil interface, and a Born approximation to model the weak contrast backscattering from buried objects. In order to address both modeling errors and ill-posedness, the proposed image reconstruction algorithms use regularization based on a total variation norm with orientation preference. The algorithms are tested on data generated by nonlinear finite difference time domain electromagnetic simulations.

Unified Anomaly Suppression and Boundary Extraction in Laser Radar Range Imagery based on a Joint Curve-Evolution and Expectation-Maximization Algorithm

In this paper, we develop a new unified approach for laser radar range anomaly suppression, range profiling, and segmentation. This approach combines an object-based hybrid scene model for representing the range distribution of the field and a statistical mixture model for the range data measurement noise. The image segmentation problem is formulated as a minimization problem which jointly estimates the target boundary together with the target region range variation and background range variation directly from the noisy and anomaly-filled range data. This formulation allows direct incorporation of prior information concerning the target boundary, target ranges, and background ranges into an optimal reconstruction process. Curve evolution techniques and a generalized expectation-maximization algorithm are jointly employed as an efficient solver for minimizing the objective energy, resulting in a coupled pair of object and intensity optimization tasks. The method directly and optimally extracts the target boundary, avoiding a suboptimal two-step process involving image smoothing followed by boundary extraction. Experiments are presented demonstrating that the proposed approach is robust to anomalous pixels (missing data) and capable of producing accurate estimation of the target boundary and range values from noisy data.

An Object-Based Contrast Source Inversion Method for Homogeneous Targets

An object-based inverse scattering algorithm is presented for electromagnetic imaging of homogeneous dielectric targets in a lossless, homogeneous background. The proposed approach embodies the use of a contrast source inversion method in conjunction with a curve-evolution-based reconstruction technique, thereby integrating the attractive computational features of the former with the robustness and edge-preserving capabilities of the latter. Numerical results involving single- and double-target configurations are presented to validate the approach and demonstrate its capabilities.

LADAR Range Image Segmentation using Curve Evolution and Maximum Likelihood Estimation

In this paper, we develop a new maximum likelihood-based, curve evolution approach for laser radar range image segmentation. This approach combines a hybrid scene model for representing the range distribution of the field and a statistical mixture model for the range data measurement noise. The image segmentation problem is formulated as an energy minimization problem which jointly estimates the target boundary together with the target region intensity and background texture directly from the noisy range data. Curve evolution techniques and an expectation-maximization algorithm are jointly employed as an efficient solver for minimizing the objective energy

Quasi-ray Gaussian beam algorithms for subsurface sensing in the presence of a moderately rough air-soil interface

A major source of variability in ground penetrating radar (GPR) interrogating signals is related to reflection from, and (double) transmission through, the rough unknown air-soil interface which obscures the useful target-scattered signals. These considerations motivated our recent investigations toward a more robust, physics-based, adaptive approach to subsurface imaging in the presence of a moderately (both in height and slope) rough air-soil interface. The proposed approach entails prior estimation of the deterministic gross features of the rough interface and subsequent exploitation of this information to enhance underground target imaging. So far, this approach has been applied to two-dimensional (2D) frequency-stepped and pulsed GPR configurations, for slightly lossy soils and low-contrast mine-like targets, yielding encouraging results. In this paper, we briefly review the proposed framework and discuss some representative results.

A statistical approach to rough surface underground imaging

We develop a statistical approach for imaging weak-contrast objects buried under a rough soil/air interface using data from a ground penetrating radar array. The imaging system uses short-pulse electromagnetic plane waves as illumination signals. We develop an approximate physical model based on a flat soil/air interface and weak-scattering to relate the collected data to the buried objects. This model uses ray optics to represent the soil/air interface, and a Born approximation to model the weak contrast backscattering from the buried objects. In order to form images of objects buried under rough soil/air interfaces, we embed our model in a statistical framework to account for modeling errors and ill-posedness. The statistical framework leads to optimal reconstruction algorithms for buried objects. We evaluate the performance of the reconstruction algorithms for different soil/air interface realizations using data generated from detailed nonlinear electromagnetic simulations of ground penetrating radar.