We present 2.5D fast and rigorous forward inversion algorithms for deep electromagnetic (EM) applications that include crosswell controlled-source EM measurements. The algorithm is based on a finite-difference approach in which multifrontal LU decomposition simulates multisource experiments at nearly the cost of simulating one single-source experiment each frequency operation. When size linear system equations large, use this noniterative solver impractical. Hence, we optimal grid technique...

In this paper, the recently developed multiplicative regularized contrast source inversion method is applied to microwave biomedical applications. The fully iterative and avoids solving any forward problem in each step. way, inverse scattering can efficiently be solved. Moreover, regularizer allows us apply blindly experimental data. We demonstrate from data collected by a 2.33-GHz circular scanner using two-dimensional (2-D) TM polarization measurement setup. Further some results of...

We discuss the problem of reconstruction profile an inhomogeneous object from scattered field data. Our starting point is contrast source inversion method, where unknown sources and are updated by iterative minimization a cost functional. possibility presence local minima nonlinear functional under which conditions they can exist. Inspired successful implementation total variation other edgepreserving algorithms in image restoration inverse scattering, we have explored use these...

In this paper, we developed a general framework for the inversion of electromagnetic measurements in cases where parametrization unknown configuration is possible.Due to ill-posed nature nonlinear inverse scattering problem, approach needed when available measurement data are limited and only carried out from transmitter-receiver positions (limited diversity).By carrying parametrization, number model parameters that need be inverted manageable.Hence Newton based can advantageously used over...

We discuss the problem of reconstruction profile a bounded object from scattered field data. Inspired by successful implementation minimization total variation (TV) in modified gradient method, we have explored possibilities this image-enhancement technique contrast source inversion (CSI) method. In order to be able implement additional regularizer CSI updating has been modified. present preconditioned conjugate method update contrast, which introduces hardly any computation time, but...

Microwave tomography is an imaging modality based on differentiation of dielectric properties object. The biological tissues and its functional changes have high medical significance. Biomedical applications microwave are a very complicated challenging problem, from both technical image reconstruction point-of-views. contrast in tissue presenting significant advantage for diagnostic purposes possesses problem image-reconstruction prospective. Different approaches been developed to attack the...

We have developed a frequency-domain joint electromagnetic (EM) and seismic inversion algorithm for reservoir evaluation exploration applications. EM data are jointly inverted using cross-gradient constraint that enforces structural similarity between the conductivity image compressional wave (P-wave) velocity image. The is based on Gauss-Newton optimization approach. Because of ill-posed nature inverse problem, regularization used to constrain solution. multiplicative technique selects...

We present a simultaneous multifrequency inversion approach for seismic data interpretation. This algorithm inverts all frequency components simultaneously. A data-weighting scheme balances the contributions from different so process does not become dominated by high-frequency components, which produce velocity image with many artifacts. Gauss-Newton minimization achieves high convergence rate and an accurate reconstructed image. By introducing modified adjoint formulation, we can calculate...

In mymargin this communication, we study the application of supervised descent method (SDM) for 2-D microwave imaging. SDM contains offline training and online prediction. stage, a data set is generated according to prior information. Then, average directions between fixed initial model models can be learned by iterative schemes. reconstruction achieved through iterations based on directions. This scheme offers new perspective incorporate information into inversion reduce computational...

Accurate determination of reservoir petrophysical parameters is great importance for monitoring and characterization. We developed a joint inversion approach the direct estimation in situ such as porosity fluid saturations by jointly inverting electromagnetic full-waveform seismic measurements. Full-waveform inversions allow exploitation full content data so that more accurate geophysical model can be inferred. Electromagnetic are linked to through Archie’s equations, whereas them...

Depicting geologic sequences from 3D seismic surveying is of significant value to subsurface reservoir exploration, but it usually time- and labor-intensive for manual interpretation by experienced interpreters. We have developed a semisupervised workflow efficient stratigraphy using the state-of-the-art deep convolutional neural networks (CNNs). Specifically, consists two components: (1) feature self-learning (SFSL) (2) model building (SMB), each which formulated as CNN. Whereas SMB...

In this paper we discuss a new type of regularization technique for the nonlinear inverse scattering problem, namely multiplicative technique. The main advantage is that do not have to determine parameter before inversion process started. We consider different norms total variation as factor. Specifically, investigate weighted L 2 ‐norm, and by using an appropriate updating scheme show factor does increase nonlinearity problem. Numerical examples synthetic experimental data demonstrate...

Low-frequency surface electromagnetic prospecting methods have been gaining a lot of interest because their capabilities to directly detect hydrocarbon reservoirs and compliment seismic measurements for geophysical exploration applications. There are two types surveys. The first is an active measurement where we use electric dipole source towed by ship over array seafloor receivers. This called the controlled-source (CSEM) method. second Magnetotelluric (MT) method driven natural sources....

We present a contrast source inversion (CSI) technique which is based on finite-difference (FD) solver for use in microwave biomedical imaging. The algorithm capable of inverting complex-permittivity data sets without the explicit forward at each iteration. FD frequency domain, utilizes perfectly matched layer (PML) boundary conditions, and stiffness matrix solved via an LU decomposition Gaussian elimination. An important feature FD-CSI that associated with depends only upon background...

We present a contrast source inversion (CSI) algorithm using finite-difference (FD) approach as its backbone for reconstructing the unknown material properties of inhomogeneous objects embedded in known background medium. Unlike CSI method integral equation (IE) approach, FD-CSI can readily employ an arbitrary medium background. The ability to use has made this very suitable be used through-wall imaging and time-lapse applications. Similar IE-CSI sources function are updated alternately...

We present a contrast source inversion (CSI) algorithm using truncated wavelet representations. Specifically, we represent the unknown sources and function in terms of basis functions. In order to reduce number coefficients for these unknowns, apply progressive multiscale truncation scheme based on reconstructed function. This approach increases robustness CSI noisy or limited data, decreases computation time as well memory usage. tested wavelet-domain method both synthetic experimental...

We present joint inversion approaches for integrating controlled source electromagnetic data and seismic full-waveform geophysical applications. The first approach is the petrophysical carried out by reconstructing parameters such as porosity saturations instead of usual resistivity, velocities mass density. This utilizes strong correlation between through relationships. Another that does not require a priori structural method. In this approach, employing regularization function enforcing...

We apply the so-called multiplicative regularized Gauss-Newton inversion algorithm for solving three-dimensional electromagnetic microwave inverse problems. This automatically adjusts regularization parameter and when combined with total variation type function, it can provide results excellent edge-preserving characteristics. In addition, in order to deal an extensive memory requirement method, we employ implicit Jacobian calculation scheme. By using this scheme do not have explicitly store...

Deep learning is applied to assist the joint inversion for audio-magnetotelluric and seismic travel time data. More specifically, deep residual convolutional neural networks (DRCNNs) are designed learn both structural similarity resistivity-velocity relationships according prior knowledge. During inversion, unknown resistivity velocity updated alternatingly with Gauss-Newton method, based on reference model generated by trained DRCNNs. The workflow of this scheme design DRCNNs explained in...

In this work, we design an iterative deep neural network to solve full-wave inverse scattering problems (ISPs) in the 2-D case. Forward modeling networks that predict scattered field are embedded inversion network. manner, predicts model update from residual between simulated data and observed data. The proposed can achieve super-resolution reconstruction meanwhile keeping of reconstructed models well consistent with We validate method both synthetic experimental inversion. Results show high...

In this study we propose a multiplicative regularization scheme to deal with the problem of detection and imaging homogeneous dielectric objects (the so-called binary objects). By considering regularizer as constraint for contrast source inversion (CSI) method are able avoid necessity determining parameter before process has been started. We present some numerical results representative two-dimensional configurations, but also show three-dimensional reconstruction full vectorial...

We have applied the finite-difference contrast-source inversion (FDCSI) method to seismic full-waveform problems. The FDCSI is an iterative nonlinear algorithm. However, unlike conjugate gradient and Gauss-Newton method, does not solve any full forward problem explicitly in each step of process. This feature makes very efficient solving large-scale computational It shown that FDCSI, with a significant lower computation cost, can produce results comparable quality those produced by better...

We present preconditioned non-linear conjugate gradient algorithms as alternatives to the Gauss-Newton method for frequency domain full-waveform seismic inversion. designed two preconditioning operators. For first preconditioner, we introduce inverse of an approximate sparse Hessian matrix. The matrix, which is highly sparse, constructed by judiciously truncating matrix based on examining auto-correlation and cross-correlation Jacobian As second employ approximation This preconditioner...