Electronique
Unsupervised weld defect classification in radiographic images using multivariate generalized Gaussian mixture model with exact computation of mean and shape parameters
In industry, the welding inspection is considered as a mandatory stage in the process of quality assurance/quality control. This inspection should satisfy the requirements of the standards and codes governing the manufacturing process in order to prevent unfair harm to the industrial plant in construction. For this purpose, in this paper, a software specially conceived for computer-aided diagnosis in weld radiographic testing is presented, where a succession of operations of preprocessing, image segmentation, feature extraction and finally defects classification is carried out on radiographic images. The last operation which is the main contribution in this paper consists in an unsupervised classifier based on a finite mixture model using the multivariate generalized Gaussian distribution (MGGD). This classifier is newly applied on a dataset of weld defect radiographic images. The parameters of the nonzero-mean MGGD-based mixture model are estimated using the Expectation-Maximization algorithm where, exact computations of mean and shape parameters are originally provided. The weld defect database represent four weld defect types (crack, lack of penetration, porosity and solid inclusion) which are indexed by a shape geometric descriptor composed of geometric measures. An outstanding performance of the proposed mixture model, compared to the one using the multivariate Gaussian distribution, is shown, where the classification rate is improved by 3.2% for the whole database, to reach more than 96%. The efficiency of the proposed classifier is mainly due to the flexible fitting of the input data, thanks to the MGGD shape parameter. Voir les détails
Mots clés : Mixture model, Multivariate GGD, radiography, weld defect, classification
Fast Adapting Mixture Parameters Schemes for Probability Density Difference-Based Deformable Model
This paper presents a new region-driven active contour using the pdf difference to evolve. The pdf estimation is done via a new and fast Gaussian mixture model (GMM) parameters updating scheme. The experiments performed on synthetic and X-ray images have shown not only an accurate contour delineation but also outstanding performance in terms of execution speed compared to the GMM estimation based on EM algorithm and to non-parametric pdf estimations. Voir les détails
Mots clés : Active contour, Adaptive mixture, GMM parameters update
Contribution à l’amélioration des performances du codage turbo dans les systèmes de transmission numériques
Pour bénéficier des propriétés des codes LDPC (Low-Density-Parity-Check) et Turbo Convolutional Codes (TCC), nous proposons un codage concaténé de type Gallager/Convolutionnel codé de la manière turbo. Le code modifié crée un équilibre entre les avantages et les inconvénients de LDPC et TCC en termes de complexité globale et de latence. Cela se fera à travers deux décodeurs SISO différents; LDPC et code convolutif récursif systématique (RSC) du même taux de code R= 1/2 sans entrelaceur. Étant donné que les deux décodeurs SISO sont de natures différentes, ils échangent des informations extrinsèques qui seront facilement adaptées l’une à l’autre. L'étude de la complexité de calcul et des performances de décodage sur un canal AWGN indique qu'une telle approche conduit à d'excellentes performances en raison de plusieurs facteurs. L'approche proposée réalise un compromis entre les régions de convergence et de plancher d'erreur. Il réduit la complexité de décodage par rapport au TCC et au 3D-TCC. Il fournit un meilleur gain de codage sur LDPC et PCGC (Parallel Concatenated Gallager Codes). Ces caractéristiques assureront un rapport coût-performance optimal. Comme ils peuvent être un meilleur choix pour les systèmes de communication d'aujourd'hui. Voir les détails
Mots clés : Complexité de calcul; Code convolutif; Information extrinsèque; LDPC; Concaténation parallèle; Turbo code.
A Bayesian Mumford–Shah Model for Radiography ImageSegmentation
This paper investigates the segmentation of radiographic images using a level set method based on a BayesianMumford–Shahmodel. The objective is to separate regions in an image that have very close arithmetic means, where a model based on thestatistical mean is not effective. Experimental results show that the proposed model can successfully separate such regions,in both synthetic images and real radiography images. Voir les détails
Mots clés : Level set
Performance of some Variational ImplicitDeformable Models on Segmenting OpticalMicroscopy Images
Industrial micrographs are used to evaluate a steelsor alloys. This assessment consists of visualizing and describingthe basic element (at the nanoscale) constituting the material.The information provided by the micrographic images needto be highlighted by image processing methods. In this paperthe performance of some region-based variational models arepresented. Such study allows to choose best models that give themore accuracy segmentation in less processing time. Voir les détails
Mots clés : Material microstructures, Microscopy images, segmentation, deformable models, region-based active contours
Study and modelling of a microwave sensor tocharacterize a dielectric materials and for CNDapplications
Non-destructive testing is a science of evaluationvarious properties of materials, without compromising itsusefulness and use. These properties can be physical, chemical,mechanical or geometrical. There are several techniques of nondestructive testing such as: acoustic emission, penetrate testing,eddy current, ultrasound and radiography, ... . However, each ofthese methods has certain limitations and disadvantages. Sincethe 1970s, some researchers have tried to use microwavetechniques to detect possible surface cracks in metal components,volumic cracks in dielectric materials or to characterize adielectric material.The objective of this article is to present a method ofcharacterization of dielectric materials, by modeling a microwavesensor. A change in the resonant frequency of the microwavesensor resulting from a change in its effective dielectric constantis considered as an index to define the dielectric constant of thesample. This work was devoted to study, modelling andrealization of a micro strip structure by the method of moments,later this structure will be simulated by a numerical modellingsoftware HFSS (High-Frequency Structure Simulator) to confirmthe results and validate the model. Voir les détails
Mots clés : Non-Destructive testing, Microwave Techniques, HFSS, Microwave Sensor, Dielectric Constant, Micro strip Modelling, moment method
Video Processing Software-based Pipeline Endoscopic Inspection
Currently, all the codes and the standards of the fluids transport industries require rigorous pipeline inspection, in order to detect all defects and anomalies and avoid leaks and failures. For this reason, a team within the division of Signal Processing and Imagery had as mission to develop an endoscope which can replace the operator inspection inside the pipeline and improve its quality and diagnostic. This endoscope named 'Pipe Explorer' is controlled by FPGA microcontrollers, and is equipped with a camera. While moving inside the pipe, the camera records a video on the memory card. In this way and in order to offer a practical tool to the operator, we have developed graphical software based on processing techniques of the stored video consisting in video preprocessing and segmentation. At the end of this processing, we obtain a video result on which appears the analysis and the interpretation of the original video to give an internal pipe quality diagnosis. The results shows all the defective areas such as corrosion which are stained with {green, blue, red} color according to its degree of severity and the risk of harmfulness on the inspected pipeline. Voir les détails
Mots clés : Pipeline inspection, endoscopy, video processing, video segmentation, corrosion.
Spatially Varying Weighting Function-BasedGlobal and Local Statistical Active Contours.Application to X-Ray Images
Image segmentation is a crucial task in the image processingfield. This paper presents a new region-based active contour whichhandles global information as well as local one, both based on the pixelsintensities. The trade-off between these information is achieved by aspatially varying function computed for each contour node location. Theapplication preliminary results of this method on computed tomographyand X-ray images show outstanding and efficient object extraction Voir les détails
Mots clés : image segmentation, Active contour, Averaged Shifted, histogram, pressure forces, statistics, Spatially varying function
Micrographic Image Segmentation using Level SetModel based on Possibilistic C-MeansClustering
Image segmentation is often required as afundamental stage in microstructure material characterization.The objective of this work is to choose hybridization betweenthe Level Set method and the clustering approach in order toextract the characteristics of the materials from thesegmentation result of the micrographic images. Morespecifically, the proposed approach contains two successivenecessary stages. The first one consists in the application ofpossibilistic c-means clustering approach (PCM) to get thevarious classes of the original image. The second stage isbased on using the result of the clustering approach i.e. oneclass among the three existing classes (which interests us) asan initial contour of the level set method to extract theboundaries of interest region. The main purpose of using theresult of the PCM algorithm as initial step of the level setmethod is to enhance and facilitate the work of the latter. Ourexperimental results on real micrographic images show thatthe proposed segmentation method can extract successfully theinterest region according to the chosen class and confirm itsefficiency for segmenting micrographic images of materials. Voir les détails
Mots clés : Level set, clustering approach, micrographic images;, image segmentation.
Robust fuzzy c-means clustering algorithmusing non-parametric Bayesian estimation in wavelet transform domain for noisy MR brain image segmentation
The major drawback of the fuzzy c-means (FCM) algorithm is its sensitivity to noise. The authors propose a new extended FCM algorithm based a non-parametric Bayesian estimation in the wavelet transform domain for segmenting noisy MR brain images. They use the Bayesian estimator to process the noisy wavelet coefficients. Before segmentation based on FCM algorithm, they use an a priori statistical model adapted to the modelisation of the wavelet coefficients of a noisy image.The main objective of this wavelet-based Bayesian statistical estimation is to recover a good quality image, from a noisy imageof poor quality. Experimental results on simulated and real magnetic resonance imaging brain images show that their proposed method solves the problem of sensitivity to noise and offers a very good performance that out performs some FCM-based algorithms. Voir les détails
Mots clés : fuzzy C-means algorithm, Non-Parametric Bayesian Estimation, Wavelet transform, image segmentation, MR Brain Images