Reza Firoozabadi, Eric L. Miller, Carey M. Rappaport and Ann W

Slides:



Advertisements
Similar presentations
Traditional practice separates seismic data processing and further interpretation. However the most efficient processing methods utilize a-priori information.
Advertisements

Steady-state heat conduction on triangulated planar domain May, 2002
Sedan Interior Acoustics
CONICAL ELECTROMAGNETIC WAVES DIFFRACTION FROM SASTRUGI TYPE SURFACES OF LAYERED SNOW DUNES ON GREENLAND ICE SHEETS IN PASSIVE MICROWAVE REMOTE SENSING.
Study of propagative and radiative behavior of printed dielectric structures using the finite difference time domain method (FDTD) Università “La Sapienza”,
2010 SKA Africa Bursary Conference Chalmers University of Technology Jian Yang, Associate Professor Chalmers University of Technology Sweden.
The 3D FDTD Buried Object Detection Forward Model used in this project was developed by Panos Kosmas and Dr. Carey Rappaport of Northeastern University.
MEASURES OF POST-PROCESSING THE HUMAN BODY RESPONSE TO TRANSIENT FIELDS Dragan Poljak Department of Electronics, University of Split R.Boskovica bb,
Image Denoising using Locally Learned Dictionaries Priyam Chatterjee Peyman Milanfar Dept. of Electrical Engineering University of California, Santa Cruz.
Millimeter Wave Sensor: An Overview
Aspects of Conditional Simulation and estimation of hydraulic conductivity in coastal aquifers" Luit Jan Slooten.
The Improved 3D Matlab_based FDFD Model and Its Application Qiuzhao Dong(NU), Carey Rapapport(NU) (contact: This.
Volkan Cevher, Marco F. Duarte, and Richard G. Baraniuk European Signal Processing Conference 2008.
Advancing Computational Science Research for Accelerator Design and Optimization Accelerator Science and Technology - SLAC, LBNL, LLNL, SNL, UT Austin,
Finite Difference Time Domain Method (FDTD)
9. Radiation & Antennas Applied EM by Ulaby, Michielssen and Ravaioli.
Interval-based Inverse Problems with Uncertainties Francesco Fedele 1,2 and Rafi L. Muhanna 1 1 School of Civil and Environmental Engineering 2 School.
Fig. 2: Radiometric angular response from deciduous Paulownia trees is plotted. The red, blue, black, and green curves trace the simulated values of four.
Implementation of 2D FDTD
Introduction to Adaptive Digital Filters Algorithms
Parameter selection in prostate IMRT Renzhi Lu, Richard J. Radke 1, Andrew Jackson 2 Rensselaer Polytechnic Institute 1,Memorial Sloan-Kettering Cancer.
1 Hybrid methods for solving large-scale parameter estimation problems Carlos A. Quintero 1 Miguel Argáez 1 Hector Klie 2 Leticia Velázquez 1 Mary Wheeler.
Chapter 15 Modeling of Data. Statistics of Data Mean (or average): Variance: Median: a value x j such that half of the data are bigger than it, and half.
1 EEE 498/598 Overview of Electrical Engineering Lecture 11: Electromagnetic Power Flow; Reflection And Transmission Of Normally and Obliquely Incident.
Adjoint Method and Multiple-Frequency Reconstruction Qianqian Fang Thayer School of Engineering Dartmouth College Hanover, NH Thanks to Paul Meaney,
Qiuzhao Dong(NU), Carey Rappapport(NU) (contact: This work was supported in part by CenSSIS, the Center for Subsurface.
INVASIVE MICROWAVE MEASUREMENT OF SOIL ELECTROMAGNETIC PROPERTIES AND CORRELATION WITH PHYSICAL QUANTITIES K. Clay, M. Farid, A. N. Alshawabkeh, C. M.
LTSI (1) Faculty of Mech. & Elec. Engineering, University AL-Baath, Syria Ahmad Karfoul (1), Julie Coloigner (2,3), Laurent Albera (2,3), Pierre Comon.
Roughness Model of Radar Backscattering From Bare Soil Surfaces Amimul Ehsan Electrical Engineering and Computer Science Department, University of Kansas.
Remarks: 1.When Newton’s method is implemented has second order information while Gauss-Newton use only first order information. 2.The only differences.
Analysis and Design of Multi-Wave Dilectrometer (MWD) for Characterization of Planetary Subsurface Using Finite Element Method Manohar D. Deshpande and.
Istanbul Technical University Electromagnetic Research Group
MULTI-DISCIPLINARY INVERSE DESIGN George S. Dulikravich Dept. of Mechanical and Aerospace Eng. The University of Texas at Arlington
1 Complex Images k’k’ k”k” k0k0 -k0-k0 branch cut   k 0 pole C1C1 C0C0 from the Sommerfeld identity, the complex exponentials must be a function.
The Perfectly Matched Layer (PML)
Research Overview Bahaa Saleh Carey Rappaport David Castañón Badri Roysam Miguel Velez-Reyes David Kaeli Research Overview Bahaa Saleh Carey Rappaport.
Enhancing One‐Dimensional FDTD
Silesian University of Technology in Gliwice Inverse approach for identification of the shrinkage gap thermal resistance in continuous casting of metals.
REMOTE SENSING AND SUBSURFACE IMAGING OF NEAR SURFACE ENVIRONMENTAL HAZARDS Student Researcher: Oluomachukwu Agwai Faculty Mentor: Reginald Eze, PhD and.
Wang Chen, Dr. Miriam Leeser, Dr. Carey Rappaport Goal Speedup 3D Finite-Difference Time-Domain.
1 The University of Mississippi Department of Electrical Engineering Center of Applied Electromagnetic Systems Research (CAESR) Atef Z. Elsherbeni
We have recently implemented a microwave imaging algorithm which incorporated scalar 3D wave propagation while reconstructing a 2D dielectric property.
Modeling Electromagnetic Fields in Strongly Inhomogeneous Media
1 EEE 431 Computational Methods in Electrodynamics Lecture 7 By Dr. Rasime Uyguroglu
The current density at each interfacial layer. The forward voltage is continuous at every point inside the body. A Layered Model for Breasts in Electrical.
Camera calibration from multiple view of a 2D object, using a global non linear minimization method Computer Engineering YOO GWI HYEON.
S Fernando Quivira, Jose Martinez Lorenzo and Carey Rappaport Gordon Center for Subsurface Sensing & Imaging Systems Northeastern University, Boston, MA.
Geometric Camera Calibration
Bounded Nonlinear Optimization to Fit a Model of Acoustic Foams
CSCE 441: Computer Graphics Forward/Inverse kinematics
UPB / ETTI O.DROSU Electrical Engineering 2
FDTD 1D-MAP Plane Wave TFSF Simulation for Lossy and Stratified Media
Radio Coverage Prediction in Picocell Indoor Networks
Ground Penetrating Radar using Electromagnetic Models
Camera Calibration Using Neural Network for Image-Based Soil Deformation Measurement Systems Zhao, Honghua Ge, Louis Civil, Architectural, and Environmental.
T. Chernyakova, A. Aberdam, E. Bar-Ilan, Y. C. Eldar
Sahar Sargheini, Alberto Paganini, Ralf Hiptmair, Christian Hafner
Non-linear Least-Squares
Accuracy of the internal multiple prediction when the angle constraints method is applied to the ISS internal multiple attenuation algorithm. Hichem Ayadi.
CSCE 441: Computer Graphics Forward/Inverse kinematics
Camera Calibration Using Neural Network for Image-Based Soil Deformation Measurement Systems Zhao, Honghua Ge, Louis Civil, Architectural, and Environmental.
Structure from Motion with Non-linear Least Squares
Haiyan Zhang and Arthur B. Weglein
Wireless Communications Chapter 4
Some remarks on the leading order imaging series
Presentation by DR. Assoc. prof. VASIL TABATADZE
A Direct Numerical Imaging Method for Point and Extended Targets
Full-waveform Inversion of GPR Data and its Frequency
Assessment of Embryo Health by Microscopy
Structure from Motion with Non-linear Least Squares
Presentation transcript:

Estimation of geometry and complex permittivity of the buried object and the medium Reza Firoozabadi, Eric L. Miller, Carey M. Rappaport and Ann W. Morgenthaler Contact: {rfirooza,elmiller,rappaport}@ece.neu.edu Boundary Definition by B-Spline: Reconstruction of dielectric properties: Dielectric Properties: Reconstruction of the geometry and electric properties of the subsurface objects from the noisy scattered field data has been a challenging problem in different environmental and biomedical research areas. In this work, we propose a method to solve this category of problems which utilizes an iterative inversion algorithm to reconstruct the shape of the object and the rough surface as well the electric parameters of the lossy object and the half-space it is buried inside. The boundaries are modeled by 2-D parametric B-spline curves determined by control points. The object and the ground are lossy and characterized by their complex permittivities. Ground is dispersive and we use single-pole conductivity model to determine its complex permittivity in each required frequency. The forward model exploited in the algorithm is SAMM (Semi-Analytic Mode Matching) which is a fast and efficient method to determine the scattered electromagnetic fields from buried objects in half space with rough interface. As an efficient numerical method, Levenberg-Marquardt algorithm is used to update the unknowns in an iterative routine coded in Matlab. The numerical experiments validate the accuracy and reliability of our inverse method. Abstract: Dispersive ground: Number of control points Basis functions of order k Control points @ ² g a v = b i ! ( 1 + z ¡ ) 2 ¢ Soil complex permittivity: Knot vector: T={t0,t1,…,tm} ² g ( ! ) = a v + i ¾ Basis Functions Final curve Single-pole conductivity model: ¾ g = b + 1 z ¡ 2 a Estimated εm= (3.5 + 0.0027i) ε0 Conclusions: Reconstruction of the buried object and the rough ground surfaces and permittivity and conductivity of soil and object was performed from observed electromagnetic field data by: Modeling the object boundary by a low-order B-spline model. Modeling the soil conductivity by single-pole conductivity model Using SAMM forward model to determine scattered field data at the location of the receivers. Using Levenberg-Marquardt iterative algorithm to solve the nonlinear least-squares optimization problem. z = e ¡ i ! ¢ T Non-dispersive object: Object complex permittivity: @ ² t r = i ² t = r + i The Forward Model: C and C’: Coordinate scattering centers, C0: Global coordinate center, I : incident plane wave, R : reflected plane wave, T : transmitted plane wave, t: cylindrical modes in object, q: cylindrical modes in ground, r: cylindrical modes in air Modes and Scattering Centers: C C’ Object Air Ground I R T t q r C0 x y z SAMM (Semi-analytical Mode Matching) forward model computes the scattered fields at receivers by enforcing boundary conditions at surfaces and computing the cylindrical rescattering modes from scattering centers. Field vector at receivers Matrix including Bessel and Hankel functions Coefficient vector: Boundary mismatch vector Problem Geometry: Unknown vector: Lossy dielectric object and lossy dispersive ground Multi-frequency incident uniform plane wave at desired angle Receivers above the interface detect the scattered fields. u = £ 1 T ; 2 g t ¤ Future Work: Simulation in 3D space. Simulation for multiple objects. Developing regularization procedures. Defining interface boundary in other compact forms. Ground Object Air Receivers Rough Interface Incident Plane Wave x y z u 1 = h p ( ) y ; 2 : N c i T x g [ ² a v b ] t r C. Boor, A Practical Guide to Splines, New York: Springer-Verlag, 1978. A. W. Morgenthaler, and C. M. Rappaport, “Scattering from dielectric objects buried beneath randomly rough ground: Validating the semi-analytic mode matching algorithm with 2-D FDFD,“ IEEE Trans. Geoscience and Remote Sensing, vol. 39, pp. 2421-2428, Nov. 2001. K. Levenberg, “A method for the solution of certain problems in least squares," Quart. Appl. Math., vol. 2, pp. 164-168,1944. C. T. Kelley, Iterative Methods for Optimization, SIAM, 1999. C. Rappaport, S. Wu, and S. Winton, “FDTD wave propagation in dispersive soil using a single pole conductivity model," IEEE Trans. Magn.,vol. 35, pp. 1542–1545, May 1999. References: Initial guess of control vector u0 Call forward solver (2-D SAMM) to Compute scattered field vector f(u) Compute cost function e (u) = || f(u) – f0 ­|| Compute new control vector u by Levenberg-Marquardt algorithm Desired accuracy achieved? Generate the surface by new control vector u No Yes Numerical Results: Example specifications Source fields: multi-frequency TM plane wave. Incidence angle: -45 degrees. SNR: 30 db. Frequency range: 300,400, … ,900 MHz. Object permittivity: εm= (2.9+0.0029i)ε0. Ground: Bosnian soil with density=1.263 g/cc, moisture level=25.3% (εav=5.03815, a1=-.925, b0=1.76106, b1=-.32102, b2=1.56193) Interface: 80 points, 11 control points. Object: 30 points, 4 control points. Publications:* R. Firoozabadi, E. L. Miller, C. M. Rappaport and A. W. Morgenthaler, “New Inverse Method for Simultaneous Reconstruction of Object Buried Beneath Rough Ground and the Ground Surface Structure Using SAMM Forward Model", IS&T/SPIE 17th Annual Symposium on Electronic Imaging: Science and Technology, pp. 382-393, San Jose, CA, 16-20 Jan. 2005. R. Firoozabadi, E. L. Miller, C. M. Rappaport and A. W. Morgenthaler, “Characterization of the Object Buried Beneath a Random Rough Ground Using a New Semi-Analytical Mode Matching Inverse Method,” IEEE AP-S International Symposium and USNC/URSI National Radio Science Meeting, Washington, DC, 3-8 July 2005. R. Firoozabadi, E. L. Miller, C. M. Rappaport and A. W. Morgenthaler, “Subsurface Estimation of the Geometry and Electromagnetic Properties of Buried Anomaly and Half-space Background with Unknown Rough Boundary”, submitted to Progress In Electromagnetics Research Symposium PIERS 2006, Cambridge, 26 - 29 March 2006. R. Firoozabadi, E. L. Miller, C. M. Rappaport and, A. W. Morgenthaler, “A New Inverse Method for Subsurface Sensing of Objects Under Randomly Rough Ground Using Scattered Electromagnetic Field Data”, to be appeared at IEEE Transactions on Geoscience and Remote Sensing. Goal: Inverse Problem Formulation: The Levenberg-Marquardt iterative algorithm is an efficient method in optimization problems in form of nonlinear least-squares minimization: Hessian : Levenberg Marquardt parameter C=I2Nc Gradient : Jacobian: The Jacobian is determined by finding the elements of: , and . Finding a good estimate to the boundaries from observed data by minimizing the cost function: Boundary reconstruction Residual vector: Predicted field Unknowns vector Observed data State of the Art: * This work was supported in part by CenSSIS, the Center for Subsurface Sensing and Imaging Systems, under the Engineering Research Centers Program of the National Science Foundation (Award Number EEC-9986821). Current forward methods: Born approximation (less accurate for large targets) FDFD & FDTD (sensitive to grid discretization and computationally expensive) SDFMM (computationally expensive) Current minimization methods: Newton Method Steepest descent method