Electronique
Inverse problem resolution using sparse regularization method based on L1 Norm. Application to X-rays tomographic reconstruction images
In this paper, a sparse regularization method in an orthogonal basis is studied and applied to X-rays tomographic reconstruction 2D images. This method is based on total variation algorithm associated to L1 norm. The inverse problem can therefore be regularized by using primal dual formulation based on proximal functions. We applied this method to non-destructive evaluation of material in the case of 2D reconstruction of X-rays tomographic images containing real defects Voir les détails
Mots clés : radiography, X Ray, L1 norm, Tomography
Optimisation et reconstruction d’images tomographiques par la méthode de la variation totale
Le problème de reconstruction d’image médical et industriel à partir des mesures de radiation effectuées autour d’un objet fait parti de la classe des problèmes inverses. La tomographie permet d’analyser et caractérisé sur écran l’intérieur des corps sans les détruire. La reconstruction d’image consiste en l’identification des propriétés interne d’un objet à partir Nous présentons le modèle de minimisation de la variation totale et ses application a la reconstruction tomographique à partir des projections d’un radioscope.Ces dernières années, La minimisation de la variation totale sous contrainte est appliquée dans plusieurs problèmes inverses mal poses en traitement d’image. Elle présente l’avantage de permettre de reconstruire des bords précis et l’implémentation est simple, elle est basée sur un algorithme de gradient. L’ajout d’information apiori est souvent facile a incorporer. Dans le domaine de l’imagerie médicale et l’astronomie, c’est souvent la cas d’utiliser la contrainte de positivité. Dans la littérature, il est souvent indiqué que la contrainte de positivité augmente la performance de l’algorithme de reconstruction. Voir les détails
Mots clés : radiography, X Ray, Monte Carlo Method
Identification parameters with neural network for Preisach hysteresis model
The description of hysteresis is one of the classical problems in magnetic materials. The progress in its solution determines the reliability of modeling and the quality of design of a wide range of devices, the proposed approach has been applied to model the behavior of many samples and the results show the robustness and efficiency of Neural Network to model the phenomenon of hysteresis loop. The goal of this study is to optimize the parameters of hysteresis Loop by Preisach model with the Neural Network, the method developed is based on an analysis of two distribution functions. The modified Lorentzian function and Gaussian function have been analyzed. The implemented software and performances of the distributions are presented. Voir les détails
Mots clés : Gaussian Distribution, Hysteresis Loop, Lorentzian Distribution, Neural Network, Preisach Model
Identification of Jiles-Atherton model parameters using genetic algorithms
The purpose of this paper is to propose a robust and fast method to estimate the parameters of Jiles-Atherton model of ferromagnetic hysteresis by using genetic algorithms for reconstruction of hysteresis loops. The performance of the method is evaluated by experimental data. Jiles-Atherton model with five parameters describes the hysteresis behavior of ferromagnetic materials. To calculate the model parameters, many researchers use analytical and classical iterative methods, the most often those methods does not converge or give unphysical results. Because the characteristic equations are sensitive to the initial values of iteration and has non-unique solutions, which is caused by nonlinearity of the characteristic equations and the fact that there are more unknown quantities. The optimization by genetic algorithms method allows to avoid local minima and find global roots is then applied to obtain the model parameters. The proposed method overcomes the difficulties of the other techniques which assume zero remanence. It is also robust as it is guaranteed to converge to physical solutions. Voir les détails
Mots clés : Jiles-Atherton model, hysteresis loops, Identification of parameters, genetic algorithms (GA)
Determination of distribution functions and optimization of parameters for the Preisach hysteresis model by Practical Swarm Optimization (PSO)
In this work we present the Preisach model for the simulation of the magnetic hysteresis. This model is able to represent the hysteresis property of versus materials if it’s distribution function µ(α,β) and his parameters are well determined. Four distributions functions, the modified Lorentzian, Guassian, Gauss-Gauss and lognormal- Gauss are used and optimized to simulate the hysteresis loops. In the last step, the optimal distribution function and there parameters of the Preisach model are determinate via practical swarm optimization (PSO) and results are compared with experimental ones. Voir les détails
Mots clés : Preisach Model, Distribution Function, practical swarm optimization (PSO)
Moment matching estimation method for an asymmetric generalized Gaussian mixture model
In this paper, the r-order moments of the asymmetric generalized Gaussian (AGG) distribution is originally computed. Then, the moment matching method associated to the expectation-maximization (EM) algorithm is used to estimate the AGG mixture model parameters. The obtained results are comparable to those of the maximum likelihood method which, however, manipulates high nonlinear equations (piece-wise function, log, etc.), contrarily to the proposed method where the calculus is less difficult. Voir les détails
Mots clés : Asymmetric generalized Gaussian distribution, finite mixture model, moment matching method, EM algorithm
Logique Floue et Traitement des Images deRadiographie Indutrielles
La segmentation est une opération très importantedans la chaîne du traitement d'images puisqu'elle contribuedirectement dans le diagnostic et la prise de décision. Dans cetravail nous proposons une approche de segmentation basée surla classification non supervisée floue en utilisant l'algorithmeFCM (fuzzy C-means), appliquée aux images de radiographie desoudure. Le but de cette segmentation est de détecter et delocaliser d’éventuels défauts qui peuvent être produits parl’opération de soudage. Voir les détails
Mots clés : Segmentation classification, étiquetage, images radiographie de soudure, contrôle non destructif, défaut de soudure
Extraction of weld defect from radiographic images using the level set segmentation without reinitialization
All level set based image segmentation methods arebased on an assumption that the level set function is close to asigned distance function (SDF). Small time step and costly reinitialization procedure must be applied to guarantee this assumption, and in order to calculate the gradient, simple numerical schemes, based on finite differences, are applied. Inthis paper, in order to achieve higher order accuracy in thetemporal discretization, we have used Total Variation Diminishing (TVD) Runge Kutta (RK) methods. The spatial derivatives are determined by using the Weighted Essentially Non-Oscillatory methods (WENO-5) that accurately capture theformation of sharp gradients in the moving fronts. In the otherhand, we have used the level set method without re-initializationin order to speed up the evolutionary process. Experiments results show that we have obtained good results both on syntheticand real images. Voir les détails
Mots clés : Image segmentation; active contour; level set
Weld Radiographic Images Segmentation and Restoration Via Local Binary Fitting Energy and Binary Level Set
The present paper is devoted to discuss a region-based active contour model formulated via level set method. Region-based models have more advantages over edge based models, especially by introducing local statistical image intensities instead of the global statistical image intensities. This propriety makes those models well adapted to segment weld radiographic images in the aim to get image with less complexity and to extract weld defects in order to use them in NDT (Non Destructive Testing). Voir les détails
Mots clés : segmentation, Restoration, weld radiographic images, region-based active contours, LBF model, Level set
Reconstruction d'image à rayon X par les algorithmes de Jacobi et Gauss-Seidel
Dans ce travail, nous présentons l’application de deux algorithmes basés sur les méthodes algébriques, de reconstruction d’image en tomographie à rayon X. Ainsi, les algorithmes de Jacobi et Gauss Seidel seront implémentés et appliqués à des images synthétiques contenant plusieurs objets et des images réels obtenues par un tomographe à Rayons X. Voir les détails
Mots clés : reconstruction, Tomographie, inversion algébrique, gauss Seidel, Jacobi