Éditeur(s) : HAL CCSD
Résumé : International audience
The fusion of regions from segmentation is often a required step in order to analyze objects in the scene. It allows to group under a same closure property the partitions of an image, which denote the same object. In this way, several studies have been made. Most of the algorithms are based on global information to evaluate the relevance of region combinations and fusion. In this paper, we propose a merging process based on Gestalt vision's theory. The particularity of our method is to preserve photographer's region of interest thanks to the definition of a new criterion based on Di Zenzo gradient. Thus, region fusion preserves photographer's relevant objects of the scene. The performance of the proposed method is evaluated using various natural images, including Berkeley image database and its accuracy is shown.
Color and Imaging Conference
San Jose, CA, United States

hal-00637144
https://hal.archives-ouvertes.fr/hal-00637144

Éditeur(s) : HAL CCSD
Résumé : Séminaire de Synthèse du Projet FPPSM "Forêts, Parcs, Pauvreté au Sud de Madagascar"
A Madagascar, la déforestation mobilise toutes les recherches, mais il manque encore une base de données complète, référencée et réactualisable sur les dynamiques forestières. Une solution est la valorisation d'images satellitaires multidates qui, de ce fait, sont issues de capteurs différents en termes de résolution spatiale, radiométrique et temporelle (Spot, Landsat, Modis). Ces données géolocalisées sont volumineuses, elles doivent être normalisées puis stockées et finalement traitées. Face à l'hétérogénéité des données et traitements, l'interopérabilité des ressources est assurée par le standard ISO (International Standard Organisation) et la spécification OGC (Open Geospatial Consortium). Le traitement de séries chronologiques de la couverture forestière exige un outil réutilisable qui s'adapte à de nouvelles données satellitaires : des chaînes de traitement pour les données simples, des workflows pour les gros volumes de données. La modélisation du workflow est illustrée par la construction de sa base de connaissance en utilisant un contexte de travail.
https://hal.archives-ouvertes.fr/hal-01370017

hal-01370017
https://hal.archives-ouvertes.fr/hal-01370017
IRD : fdi:010066424

Éditeur(s) : HAL CCSD
Séminaire du Laboratoire de Mathématiques Informatique et Applications (LAMIA)
Pointe-à-Pitre, Guadeloupe

hal-00835738
https://hal.archives-ouvertes.fr/hal-00835738

Éditeur(s) : HAL CCSD
Résumé : National audience
RFIA
Lyon, France

hal-01313397
https://hal.archives-ouvertes.fr/hal-01313397

Éditeur(s) : HAL CCSD
Résumé : International audience
SELPER
Cayenne, French Guiana

hal-01313392
https://hal.archives-ouvertes.fr/hal-01313392

Éditeur(s) : HAL CCSD
Résumé : National audience
Séminaire du Laboratoire de Mathématiques Informatique et Applications (LAMIA)
Pointe-à-Pitre, Guadeloupe

hal-00694143
https://hal.archives-ouvertes.fr/hal-00694143

Éditeur(s) : HAL CCSD
Résumé : International audience
This paper deals with a comparison of different colour space in order to improve high resolution images classification. The background of this study is the measure of the agriculture impact on the environment in islander context. Biodiversity is particularly sensitive and relevant in such areas and the follow-up of the forest front is a way to ensure its preservation. Very high resolution satellite images are used such as QuickBird and IKONOS scenes. In order to segment the images into forest and agriculture areas, we characterize both ground covers with colour and texture features. A classical unsupervised classifier is then used to obtain labelled areas. As features are computed on coloured images, we can wonder if the colour space choice is relevant. This study has been made considering more than fourteen colour spaces (RGB, YUV, Lab, YIQ, YCrCs, XYZ, CMY, LMS, HSL, KLT, IHS, I1I2I3, HSV, HSI, etc.) and shows the visual and quantitative superiority of IHS on all others. For conciseness reasons, results only show RGB, I1I2I3 and IHS colour spaces.
SPIE Remote Sensing Proceedings
Cardiff, United Kingdom

hal-00602277
https://hal.archives-ouvertes.fr/hal-00602277

Éditeur(s) : HAL CCSD Elsevier
Résumé : International audience
Voting rules relying on k-nearest neighbors (k-NN) are an effective tool in countless many machine learning techniques. Thanks to its simplicity, k-NN classification is very attractive to practitioners, as it enables very good performances in several practical applications. However, it suffers from various drawbacks, like sensitivity to "noisy" instances and poor generalization properties when dealing with sparse high-dimensional data. In this paper, we tackle the k-NN classification problem at its core by providing a novel k-NN boosting approach. Namely, we propose a supervised learning algorithm, called Universal Nearest Neighbors (UNN), that induces a leveraged k-NN rule by globally minimizing a surrogate risk upper bounding the empirical misclassification rate over training data. Interestingly, this surrogate risk can be arbitrary chosen from a class of Bregman loss functions, including the familiar exponential, logistic and squared losses. Furthermore, we show that UNN allows to efficiently filter a dataset of instances by keeping only a small fraction of data. Experimental results on the synthetic Ripley's dataset show that such a filtering strategy is able to reject "noisy" examples, and yields a classification error close to the optimal Bayes error. Experiments on standard UCI datasets show significant improvements over the current state of the art.
ISSN: 0925-2312

hal-00664462
https://hal.inria.fr/hal-00664462

Éditeur(s) : HAL CCSD
Résumé : International audience
Le nombre de satellites et de capteurs pour la télédétection dédiés à l'observation de la Terre ne cesse d'augmenter, permettant ainsi d'avoir une masse de données importante en particulier en matière d'images. Parallèlement, un effort permanent vise, d'une part, à améliorer l'accès à ces données et, d'autre part, à développer d'avantages d'outils pour les manipuler. De tels efforts sont particulièrement utiles dans des contextes socio-environnementaux très dynamiques spatialement et temporellement, pour lesquels il est nécessaire de suivre et de prédire les événements environnementaux et sociétaux. En revanche, en présence d'un tel flux de données, nous avons besoin de méthodes automatiques d'interprétation d'images. Une solution envisageable pour répondre à ce besoin est de bénéficier des atouts de l'intelligence artificielle pour l'obtention de cartes d'occupation du sol issues d'une classification des régions des images. Afin de contribuer à l'automatisation de la classification, nous proposons une méthode d'induction de règles interprétables par des non-experts et mettant en évidence, explicitement, des connaissances structurelles. Cette méthode s'appuie sur la programmation logique inductive (PLI) et en particulier sur le système inductif ”Aleph”. L'application de la méthode de classification Multi-class Rule Set Intersection (MRSI) permet ensuite de classifier tout nouvel objet au regard des ses caractéristiques intrinsèques et de celles des objets environnants. Nous avons appliqué notre méthodologie à l'étude de la dynamique du littoral de la Guyane Française. Suite à ce travail, nous avons induit 136 règles de classification pour 38 classes d'occupation du sol. Ces règles sont intelligibles et simples à interpréter de par l'utilisation de la logique du premier ordre. Les performances du système ont été évaluées par la validation croisée. En moyenne, la précision, la spécificité et la sensibilité sont, respectivement, égales à 84,62%, 99,57% et 77,22%. Ces résultats quantitatifs montrent une bonne performance de la méthodologie pour la mise à jour automatique de cartes d'occupation du sol et/ou l'assistance aux opérateurs utilisant l'analyse d'image orientée-objet.
11èmes Rencontres des Jeunes Chercheurs en Intelligence Artificielle
Lille, France

hal-01376400
https://hal.archives-ouvertes.fr/hal-01376400
https://hal.archives-ouvertes.fr/hal-01376400/document
https://hal.archives-ouvertes.fr/hal-01376400/file/Bayoudh_RJCIA2013.pdf

Éditeur(s) : HAL CCSD
Résumé : International audience
Proceedings of Geoprocessing

hal-00509489
https://hal.archives-ouvertes.fr/hal-00509489

Éditeur(s) : HAL CCSD
Résumé : International audience
In this paper, we use a segmentation method applied to the plants images analysis. The segmentation gives groups of pixels, and we propose to use merging operators to address the most relevant groups in order to improve the recognitions systems performance. In particular, we use an operator called Pi fully reinforced. The experiments carried out show that the use of Pi improves outcomes for structuring groups of pixels.
Advanced Information Networking and Applications
Fukuoka, Japan

hal-00652125
https://hal.archives-ouvertes.fr/hal-00652125
DOI : 10.1109/WAINA.2012.40

Éditeur(s) : HAL CCSD Elsevier
Résumé : International audience
Experimentally Assisted Molecular Modelling (EAMM) is an original approach for predicting the structure of organic micro-crystals and deducing their habit in the presence of various solvents and additives. It is applied here in the case of n-octylamino-nitrobenzoxadiazole (nOA-NBD), a fluorescent compound. This general approach is first of all described and validated by its blind application to three known and closely-related crystals. Then, the whole process is applied to predict the molecular crystal generated by n-octylamino-NBD, including the molecule conformation, its structure (cell parameters, molecular packing, X-ray powder diagrams) and its theoretical plus actual habits in the presence of a solvent (p-xylene) and two additives (acetic acid and n-dodecane). The conformation of the predicted molecule is validated by comparison with two similar molecules embedded in observable crystals. Then, the choice between two proposed structures (with the same energy and two equiprobable packings) is based on the comparison between experimental and re-computed X-ray powder diffraction diagrams, and also on the interpretation of an actual TEM image in the light of the proposed cell parameters. This choice is confirmed by the comparison between the vapour-grown face development (proposed by the BFDH or the attachment energy models) and the actual face development in a pure solvent. Finally, the actual habit deduced from comparing the crystal attachment energy with the adsorption energy of solvent or additive molecules is validated by SEM images of NBD micro-crystals in presence of the solvent alone or in presence of both additives. In conclusion, the EAMM approach appears to be a decisive tool for quickly simulating structural and habit properties of the molecular crystals, accessible or not by experimentation. Thus, a numerical selection of molecules becomes possible in view of deriving organic materials with predetermined properties, such as the fluorescence properties of the compound considered in the present study.
ISSN: 0022-0248

hal-00857894
https://hal.archives-ouvertes.fr/hal-00857894
DOI : 10.1016/j.jcrysgro.2011.07.035

Éditeur(s) : HAL CCSD
Résumé : International audience
The Corinth Rift is the most seismically active zone in Europe. Since the year 2000, the Corinth Rift Laboratory (CRL, http://crlab.eu) consisting in a multidisciplinary natural observatory, aims at understanding the mechanics of faulting and earthquake nucleation in the Rift. Recent studies have improved our view about fault geometry and mechanics within CRL but there is still a critical need for a better knowledge of the structure at depth.In this project, we aim to analyze the complete seismological database (13 years of recordings) of CRL by using recently developed methodologies of structural imaging in order to determine, with high resolution, the local 3D structure and the earthquake locations. We perform an iterative joint determination of 3D velocity model and earthquake coordinates. P and S velocity models are estimated through a first arrival traveltime tomography by using the adjoint state method proposed by Taillandier et al. (2009). It consists in the minimization of the cost function between observed and theoretical arrival times,which is achieved by estimating the gradient of the cost function. Earthquake locations are also estimated with a linearized inversion approach by computing the gradient of the misfit function. Iterations are repeated until the cost function no longer decreases. We present preliminary results consisting in: (1) the adjustment of a 1D velocity model that is used as initial model of the 3D tomography and (2) a first attempt of the joint determination of 3D velocity model and earthquake locations.We also perform a preliminary receiver function analysis of teleseismic data recorded by the broadband stations of the CRL network. The multiscale approach (Gesret et. al., 2010) analysis of teleseismic P to S converted waves allows to analyze the signal toward the higher frequencies and thus to characterize the structure at depth with a high resolution. Our multiscale approach should give us reliable depth of interfaces beneath seismometers in the CRL area. In this first attempt, we adjust the 1D velocity model that produces a synthetic RF as similar as possible to the stack of observed RF for a subset of data. The identified interfaces could then be compared with heterogeneities imaged by the tomography.
AGU Fall Meeting
San Fransico, United States

hal-01417184
https://hal-mines-paristech.archives-ouvertes.fr/hal-01417184

Éditeur(s) : HAL CCSD
Résumé : International audience
World's Large Rivers Conference
Manaus, Brazil

hal-01313383
https://hal.archives-ouvertes.fr/hal-01313383

Éditeur(s) : HAL CCSD
Résumé : International audience
The Corinth Rift is the most seismically active zone in Europe. The area is characterized by very localized NSextension at a rate of1.5cm/year, the occurrence of frequent and intensive microseismic crises and occasionalmoderate to large earthquakes like in 1995 (Mw=6.1).Since the year 2000, the Corinth Rift Laboratory (CRL, http://crlab.eu) consisting in a multidisciplinary naturalobservatory, aims at understanding the mechanics of faulting and earthquake nucleation in the Rift. Recent studieshave improved our view about fault geometry and mechanics within CRL, but there is still a critical need for a betterknowledge of the structure at depth both for the accuracy of earthquake locations and for mechanical interpretationof the seismicity.In this project, we aim to analyze the complete seismological database (13 years of recordings) of CRL by usingrecently developed methodologies of structural imaging, in order to determine at the same time and with highresolution, the local 3D structure and the earthquake locations.We perform an iterative joint determination of 3D velocity model and earthquake coordinates. In a first step, Pand S velocity models are determined using first arrival time tomography method proposed by Taillandier et al.(2009). It consists in the minimization of the cost function between observed and theoretical arrival times which isachieved by the steepest descent method (e.g. Tarantola 1987). This latter requires computing the gradient of thecost function by using the adjoint state method (Chavent 1974). In a second step, earthquakes are located in thenew velocity model with a non-linear inversion method based on a Bayesian formulation (Gesret et al. 2015). Step1 and 2 are repeated until the cost function no longer decreases.We present preliminary results consisting in: (1) the adjustement of a 1D velocity model that is used as initialmodel of the 3D tomography and (2) a first attempt of the joint determination of 3D velocity model and earthquakelocation.In addition to the tomographic imaging, we perform a preliminary receiver function analysis of teleseismic datarecorded by the broadband stations of the CRL network. The RF analysis should provide the interface depthsbeneath seismometers and increase the imaging resolution of the upper crustal structures provided by the 3Dtomography. In this first attempt, we adjust the 1D velocity model that produces a synthetic RF as similar aspossible to the observed RF for a subset of data. We compare the identified interfaces with structures imaged bythe tomography.
EGU general assembly 2016
Vienne, Austria

hal-01306224
https://hal-mines-paristech.archives-ouvertes.fr/hal-01306224

Éditeur(s) : HAL CCSD
Résumé : Afin de prévoir l'évolution des phénomènes naturels, sans discontinuité, des milliers d'images satellites alimentent les stations météorologiques. Ce flot permanent d'informations nécessite rapidité et rigueur d'analyse. Il est donc indispensable de mettre en place des systèmes capables de travailler nuit et jour afin de dresser des bilans exploitables par les chercheurs. Nous proposons dans cet article une méthode, basée sur des techniques d'investigation en reconnaissance des formes, telles que la segmentation et la classification. Ces techniques seront utilisées pour localiser et discriminer les masses nuageuses issues d'images captées par le satellite géostationnaire GOES. Le traitement des formes nuageuses trouvées se fait en utilisant des opérations, telles que la morphologie mathématique afin d'extraire des caractéristiques discriminantes. Pour l'interprétation des caractéristiques obtenues et pour effectuer la classification des nuages, nous utilisons des systèmes de classeurs.
Journées d'Études et de Recherche des Antilles et de la Guyane
Pointe-à-Pitre, Guadeloupe

hal-00637787
https://hal.archives-ouvertes.fr/hal-00637787

Éditeur(s) : HAL CCSD IRD
Résumé : International audience
The number of satellites and remote sensing sensors devoted to earth observation becomes increasingly high, providing more and more data and especially images. In the same time the access to such a data and to the tools to process them has been considerably improved. In the presence of such data flow - and regarding the necessity to follow up and predict environmental and societal changes in highly dynamic socio-environmental contexts - we need automatic image interpretation methods. This could be accomplished by exploring some strengths of artificial intelligence. Our main idea consists in inducing classification rules that explicitly take into account structural knowledge, using Aleph, an Inductive Logic Programming (ILP) system. We applied our proposed methodology to three land cover/use maps of the French Guiana littoral. One hundred and forty six classification rules were induced for the 39 land-cover classes of the maps. These rules are expressed in first order logic language which make them intelligible and interpretable by non-experts. A ten-fold cross validation gave average values for classification accuracy, specificity and sensibility equal to, respectively, 98.82 %, 99.65% and 70%. The proposed methodology could be valuably exploited to automatically classify new objects and/or help operators using object-based classification procedures.
Symposium of the Latin American Society for Remote Sensing and Spatial Information Systems (SELPER)
Cayenne, French Guiana

ird-00797337
http://hal.ird.fr/ird-00797337
http://hal.ird.fr/ird-00797337/document
http://hal.ird.fr/ird-00797337/file/Bayoudh_et_al_2012_FP_SELPER-086.pdf
IRD : fdi:010058343

Éditeur(s) : HAL CCSD Springer Verlag
Résumé : under revision for IJCV
International audience
The k-nearest neighbors (k-NN) classification rule has proven extremely successful in countless many computer vision applications. For example, image categorization often relies on uniform voting among the nearest prototypes in the space of descriptors. In spite of its good properties, the classic k-NN rule suffers from high variance when dealing with sparse prototype datasets in high dimensions. A few techniques have been proposed to improve k-NN classification, which rely on either deforming the nearest neighborhood relationship or modifying the input space. In this paper, we propose a novel boosting algorithm, called UNN (Universal Nearest Neighbors), which induces leveraged k-NN, thus generalizing the classic k-NN rule. We redefine the voting rule as a strong classifier that linearly combines predictions from the k closest prototypes. Weak classifiers are learned by UNN so as to minimize a surrogate risk. A major feature of UNN is the ability to learn which prototypes are the most relevant for a given class, thus allowing one for effective data reduction. Experimental results on the synthetic two-class dataset of Ripley show that such a filtering strategy is able to reject "noisy" prototypes. We carried out image categorization experiments on a database containing eight classes of natural scenes. We show that our method outperforms significantly the classic k-NN classification, while enabling significant reduction of the computational cost by means of data filtering.
ISSN: 0920-5691

hal-00481712
https://hal.archives-ouvertes.fr/hal-00481712
DOI : 10.1007/s11263-012-0539-2
ARXIV : 1001.1221

Éditeur(s) : HAL CCSD
Résumé : International audience
In this paper we propose a hybrid approach using mutual information and multi-objective optimization for feature subset selection problem. The hybrid aspect is due to the sequence of a filter method and a wrapper method in order to take advantages of both. The filter method reduces the exploration space by keeping subsets having good internal properties and the wrapper method chooses among the remaining subsets with a classification performances criterion. In the filter step, the subsets are evaluated in a multi-objective way to ensure diversity within the subsets. The evaluation is based on the mutual information to estimate the dependency between features and classes and the redundancy between features within the same subset. We kept the non-dominated (Pareto optimal) subsets for the second step. In the wrapper step, the selection is made according to the stability of the subsets regarding classification performances during learning stage on a set of classifiers to avoid the specialization of the selected subsets for a given classifiers. The proposed hybrid approach is experimented on a variety of reference data sets and compared to the classical feature selection methods FSDD and mRMR. The resulting algorithm outperforms these algorithms and the computation complexity remains acceptable even if it increases with regards to these two fast selection methods.
Journal of Informatics and Mathematical Sciences

hal-01375982
https://hal.archives-ouvertes.fr/hal-01375982

Éditeur(s) : HAL CCSD
Résumé : National audience
Dans cet article nous présentons les résultats opérationnels obtenus pour la pré-détection de bateaux de pêche dans des images satellites optiques haute résolution. Les images utilisées sont de type SPOT 5 panchromatique de résolution 5m. La méthode utilisée s'appuie sur les techniques Bayésiennes classiques de détection, et ne comporte aucun pré-traitement. Au cours de cet article, nous décrivons précisément la méthode utilisée, ainsi que le réglage des deux seuls paramètres de la méthode qui sont le seuil de détection et la taille de la fenêtre d'analyse. Ces paramètres sont fixés à partir des courbes ROC (Receiver Operating Characteristic) obtenues sur les jeux de tests dont nous disposons. La méthode est enfin testée sur un jeu complet et réaliste d'images. La méthode doit être à terme intégrée dans une application industrielle.
ORASIS'09 - Congrès des jeunes chercheurs en vision par ordinateur
Trégastel, France, France

Droits : info:eu-repo/semantics/OpenAccess
inria-00404625
https://hal.inria.fr/inria-00404625
https://hal.inria.fr/inria-00404625/document
https://hal.inria.fr/inria-00404625/file/23.pdf