|
Publications sur Segmentation
Résultat de la recherche dans la liste des publications :
5 Articles |
1 - A study of Gaussian mixture models of colour and texture features for image classification and segmentation. H. Permuter et J.M. Francos et I. H. Jermyn. Pattern Recognition, 39(4): pages 695--706, avril 2006. Mots-clés : Classification, Segmentation, Texture, Couleur, Mixture de gaussiennes, Decison fusion.
@ARTICLE{permuter_pr06,
|
author |
= |
{Permuter, H. and Francos, J.M. and Jermyn, I. H.}, |
title |
= |
{A study of Gaussian mixture models of colour and texture features for image classification and segmentation}, |
year |
= |
{2006}, |
month |
= |
{avril}, |
journal |
= |
{Pattern Recognition}, |
volume |
= |
{39}, |
number |
= |
{4}, |
pages |
= |
{695--706}, |
url |
= |
{http://dx.doi.org/10.1016/j.patcog.2005.10.028}, |
pdf |
= |
{ftp://ftp-sop.inria.fr/ariana/Articles/2006_permuter_pr06.pdf}, |
keyword |
= |
{Classification, Segmentation, Texture, Couleur, Mixture de gaussiennes, Decison fusion} |
} |
Abstract :
The aims of this paper are two-fold: to define Gaussian mixture models of coloured texture on several feature paces and to compare the performance of these models
in various classification tasks, both with each other and with other models popular in the literature. We construct Gaussian mixtures models over a variety of different colour and texture feature spaces, with a view to the retrieval of textured colour images from databases. We compare supervised classification results for different choices of colour and texture features using the Vistex database, and explore the best set of features and the best GMM configuration for this task. In addition we introduce several methods for combining the 'colour' and 'structure' information in order to improve the classification performance. We then apply the resulting models to the classification of texture databases and to the classification of man-made and natural areas in aerial images. We compare the GMM model with other models in the literature, and show an overall improvement in performance. |
|
2 - An approximation of the Mumford-Shah energy by a family of dicrete edge-preserving functionals. G. Aubert et L. Blanc-Féraud et R. March. Nonlinear Analysis, 64: pages 1908-1930, 2006. Mots-clés : Gamma Convergence, Elements finis, Segmentation.
@ARTICLE{laure-na05,
|
author |
= |
{Aubert, G. and Blanc-Féraud, L. and March, R.}, |
title |
= |
{An approximation of the Mumford-Shah energy by a family of dicrete edge-preserving functionals}, |
year |
= |
{2006}, |
journal |
= |
{Nonlinear Analysis}, |
volume |
= |
{64}, |
pages |
= |
{1908-1930}, |
pdf |
= |
{ftp://ftp-sop.inria.fr/ariana/Articles/2006_laure-na05.pdf}, |
keyword |
= |
{Gamma Convergence, Elements finis, Segmentation} |
} |
Abstract :
We show the Gamma-convergence of a family of discrete functionals to the Mumford and Shah image segmentation functional.
The functionals of the family are constructed by modifying the elliptic approximating functionals proposed by Ambrosio and Tortorelli. The quadratic term of the energy related to the edges of the segmentation is replaced by a nonconvex functional. |
|
3 - Detecting codimension-two objects in an image with Ginzburg-Landau models. G. Aubert et J.F. Aujol et L. Blanc-Féraud. International Journal of Computer Vision, 65(1-2): pages 29-42, novembre 2005. Mots-clés : Modele de Ginzburg-Landau, Detection de points, Segmentation, PDE, Images biologiques, Images SAR.
@ARTICLE{laure-ijcv05,
|
author |
= |
{Aubert, G. and Aujol, J.F. and Blanc-Féraud, L.}, |
title |
= |
{Detecting codimension-two objects in an image with Ginzburg-Landau models}, |
year |
= |
{2005}, |
month |
= |
{novembre}, |
journal |
= |
{International Journal of Computer Vision}, |
volume |
= |
{65}, |
number |
= |
{1-2}, |
pages |
= |
{29-42}, |
pdf |
= |
{ftp://ftp-sop.inria.fr/ariana/Articles/GL_IJCV_5.pdf}, |
keyword |
= |
{Modele de Ginzburg-Landau, Detection de points, Segmentation, PDE, Images biologiques, Images SAR} |
} |
Abstract :
In this paper, we propose a new mathematical model for detecting in an image singularities of codimension greater than or equal to two. This means we want to detect points in a 2-D image or points and curves in a 3-D image. We drew one's inspiration from
Ginzburg-Landau (G-L) models which have proved their efficiency for modeling many phenomena in physics. We introduce the model, state its
mathematical properties and give some experimental results demonstrating its capability in image processing. |
|
4 - Supervised Segmentation of Remote Sensing Images Based on a Tree-Structure MRF Model. G. Poggi et G. Scarpa et J. Zerubia. IEEE Trans. Geoscience and Remote Sensing, 43(8): pages 1901-1911, août 2005. Mots-clés : Classification, Segmentation, Champs de Markov.
@ARTICLE{ieeetgrs_05,
|
author |
= |
{Poggi, G. and Scarpa, G. and Zerubia, J.}, |
title |
= |
{Supervised Segmentation of Remote Sensing Images Based on a Tree-Structure MRF Model}, |
year |
= |
{2005}, |
month |
= |
{août}, |
journal |
= |
{IEEE Trans. Geoscience and Remote Sensing}, |
volume |
= |
{43}, |
number |
= |
{8}, |
pages |
= |
{1901-1911}, |
pdf |
= |
{http://ieeexplore.ieee.org/iel5/36/32001/01487647.pdf?tp=&arnumber=1487647&isnumber=32001}, |
keyword |
= |
{Classification, Segmentation, Champs de Markov} |
} |
|
5 - Globally optimal regions and boundaries as minimum ratio weight cycles. I. H. Jermyn et H. Ishikawa. IEEE Trans. Pattern Analysis and Machine Intelligence, 23(10): pages 1075-1088, octobre 2001. Mots-clés : Graphe, Ratio, Cycle, Segmentation, Minimum global. Copyright : ©2008 IEEE. Personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution to servers or lists, or to reuse any copyrighted component of this work in other works must be obtained from the IEEE.
@ARTICLE{jermyn_tpami01,
|
author |
= |
{Jermyn, I. H. and Ishikawa, H.}, |
title |
= |
{Globally optimal regions and boundaries as minimum ratio weight cycles}, |
year |
= |
{2001}, |
month |
= |
{octobre}, |
journal |
= |
{IEEE Trans. Pattern Analysis and Machine Intelligence}, |
volume |
= |
{23}, |
number |
= |
{10}, |
pages |
= |
{1075-1088}, |
url |
= |
{http://dx.doi.org/10.1109/34.954599}, |
pdf |
= |
{ftp://ftp-sop.inria.fr/ariana/Articles/jermyn_tpami01.pdf}, |
keyword |
= |
{Graphe, Ratio, Cycle, Segmentation, Minimum global} |
} |
Abstract :
We describe a new form of energy functional for the modelling and identification of regions in images. The energy is defined on the space of boundaries in the image domain, and can incorporate very general combinations of modelling information both from the boundary (intensity gradients,ldots), em and from the interior of the region (texture, homogeneity,ldots). We describe two polynomial-time digraph algorithms for finding the em global minima of this energy. One of the algorithms is completely general, minimizing the functional for any choice of modelling information. It runs in a few seconds on a 256 times 256 image. The other algorithm applies to a subclass of functionals, but has the advantage of being extremely parallelizable. Neither algorithm requires initialization. |
|
haut de la page
3 Thèses de Doctorat et Habilitations |
1 - Détection de Filaments dans des images 2D et 3D; modélisation, étude mathématique et algorithmes.. A. Baudour. Thèse de Doctorat, Universite de Nice Sophia Antipolis, mai 2009. Mots-clés : imagerie 3D, Segmentation, filaments, Deconvolution, Methodes variationnelles, mocroscopie confocale.
@PHDTHESIS{baudour2009,
|
author |
= |
{Baudour, A.}, |
title |
= |
{Détection de Filaments dans des images 2D et 3D; modélisation, étude mathématique et algorithmes.}, |
year |
= |
{2009}, |
month |
= |
{mai}, |
school |
= |
{Universite de Nice Sophia Antipolis}, |
url |
= |
{https://hal.inria.fr/tel-00507520/}, |
keyword |
= |
{imagerie 3D, Segmentation, filaments, Deconvolution, Methodes variationnelles, mocroscopie confocale} |
} |
Résumé :
Cette thèse aborde le problème de la modélisation et de la détection des laments
dans des images 3D.
Nous avons développé des méthodes variationnelles pour quatre applications
spéciques :
l'extraction de routes où nous avons introduit la notion de courbure totale
pour conserver les réseaux réguliers en tolérant les discontinuités de
direction.
la détection et la complétion de laments fortement bruités et présentant
des occultation. Nous avons utilisé la magnétostatique et la théorie
de Ginzburg-Landau pour représenter les laments comme ensemble de
singularités d'un champ vectoriel.
la détection de laments dans des images biologiques acquises en microscopie
confocale. On modélise les laments en tenant compte des spécicité
de cette dernière. Les laments sont alors obtenus par une méthode de
maximum à posteriori.
la détection de cible dans des séquences d'images infrarouges. Dans cette
application, on cherche des trajectoires optimisant la diérence de luminosit
é moyenne entre la trajectoire et son voisinage en tenant compte des
capteurs utilisés.
Par ailleurs, nous avons démontré des résultats théoriques portant sur la
courbure totale et la convergence de la méthode d'Alouges associée aux systèmes
de Ginzburg-Landau. Ce travail réunit à la fois modélisation, résulats théoriques
et recherche d'algorithmes numériques performants permettant de traiter de
réelles applications. |
|
2 - Indexing of satellite images using structural information. A. Bhattacharya. Thèse de Doctorat, Ecole Nationale Supérieure des Télécommunications, 2007. Mots-clés : Landscape, Segmentation, Features, Extraction, Classification, Data mining.
@PHDTHESIS{bhattacharya_these,
|
author |
= |
{Bhattacharya, A.}, |
title |
= |
{Indexing of satellite images using structural information}, |
year |
= |
{2007}, |
school |
= |
{Ecole Nationale Supérieure des Télécommunications}, |
pdf |
= |
{ftp://ftp-sop.inria.fr/ariana/Articles/2007_bhattacharya_these.pdf}, |
keyword |
= |
{Landscape, Segmentation, Features, Extraction, Classification, Data mining} |
} |
|
3 - Analyse de Texture par Méthodes Markoviennes et par Morphologie Mathématique : Application à l'Analyse des Zones Urbaines sur des Images Satellitales. A. Lorette. Thèse de Doctorat, Universite de Nice Sophia Antipolis, septembre 1999. Mots-clés : Texture, Segmentation, Champs de Markov, Morphologie mathematique, Zones urbaines.
@PHDTHESIS{lorette99,
|
author |
= |
{Lorette, A.}, |
title |
= |
{Analyse de Texture par Méthodes Markoviennes et par Morphologie Mathématique : Application à l'Analyse des Zones Urbaines sur des Images Satellitales}, |
year |
= |
{1999}, |
month |
= |
{septembre}, |
school |
= |
{Universite de Nice Sophia Antipolis}, |
pdf |
= |
{Theses/these-lorette.pdf}, |
keyword |
= |
{Texture, Segmentation, Champs de Markov, Morphologie mathematique, Zones urbaines} |
} |
Résumé :
Dans cette thèse, nous nous intéressons au problème de l'analyse urbaine à partir d'images satellitales par des méthodes automatiques ou semi-automatiques issues du traitement d'image. Dans le premier chapitre, nous présentons le contexte dans lequel le travail a été effectué. Nous exposons les types de données utilisées, les approches statistiques considérées. Nous donnons également quelques exemples d'applications qui justifient une telle étude. Enfin, un état de l'art des diverses méthodes d'analyse des textures est présenté. Dans les deux chapitres suivants, nous développons une méthode automatique d'extraction d'un masque urbain à partir d'une analyse de la texture de l'image. Des méthodes d'extraction d'un masque urbain sont décrites. Ensuite, nous définissons plus précisemment les huit modèles markoviens gaussiens fondés sur des chaines. Ces modèles sont renormalisés par une méthode de renormalisation de groupe issue de la physique statistique afin de corriger le biais introduit par l'anisotropie du réseau de pixels. L'analyse de texture proposée est comparée avec deux méthodes classiques: les matrices de cooccurrence et les filtres de Gabor. L'image du paramètre de texture est ensuite classifiée avec un algorithme non supervisé de classification floue fondée sur la définition d'un critère entropique. Les paramètres estimés avec cet algorithme sont intégrés dans un modèle markovien de segmentation. Des résultats d'extraction de masques urbains sont finalement présentés sur des images satellitales optiques SPOT3, des simulations SPOT5, et des images radar ERS1. Dans le quatrième chapitre, nous présentons l'analyse granulométrique utilisée pour analyser le paysage urbain. Les outils et définitions de base de la morphologie mathématique sont exposés. Nous nous intéressons plus particulièrement à l'ouverture par reconstruction qui est utilisée comme transformation de base de la granulométrie. L'étape de quantification qui suit tout étape de transformation nous permet d'estimer en chaque pixel une distribution locale de taille qui est intégrée dans le terme d'attache aux données d'un modèle markovien de segmentation. Des tests sont effectués sur des simulations SPOT5. |
Abstract :
In this thesis, we investigate the problem of urban areas analysis from satellite images by automatic or semi-automatic methods coming from image processing. In the first chapter, we describe the context of this work, i.e. the type of used data, the statistical applied methods. We also give some examples of the applications which require such an analysis. Finally, a study of the existing methods of texture analysis is presented. In the second and third chapter, we develop a non supervised method based on texture analysis in order to extract an urban mask. First a study of the existing methods of urban mask extraction is presented. Second we precisely describe the eight chain-based Gaussian Markovian models used to characterize urban texture. These models are normalized through a renormalization group technique derived from statistical physics in order to correct the bias introduced by the anisotropy of the lattice.The above mentionned method of texture analysis is then compared with two classical ones: coocurrences matrix and Gabor filters. The image is then partitionned by an unsupervised fuzzy Cmeans algorithm based on an entropic criterion. The final segmentation is performed by the minimization of an energy derived from a Markovian model. Some results are presented that are obtained from SPOT3 images, SPOT5 simulations and radar ERS1 images. In the fourth chapter, we present the granulometric approach used to segment within the urban area itself. The basic operations and definitions of mathematical morphology are settled. We are particularly interested in opening by reconstruction operation based on geodesic dilatations. In fact this operation is used to define a granulometry. The quantification step that follows the transformation step consists in estimating a local size distribution function for each pixel. These parameters are then integrated in the data term of a Markovian model. Some results on SPOT5 simulations are presented. |
|
haut de la page
10 Articles de conférence |
1 - Morphological road segmentation in urban areas from high resolution satellite images. R. Gaetano et J. Zerubia et G. Scarpa et G. Poggi. Dans International Conference on Digital Signal Processing, Corfu, Greece, juillet 2011. Mots-clés : Segmentation, Classification, skeletonization , pattern recognition, shape analysis.
@INPROCEEDINGS{GaetanoDSP,
|
author |
= |
{Gaetano, R. and Zerubia, J. and Scarpa, G. and Poggi, G.}, |
title |
= |
{Morphological road segmentation in urban areas from high resolution satellite images}, |
year |
= |
{2011}, |
month |
= |
{juillet}, |
booktitle |
= |
{International Conference on Digital Signal Processing}, |
address |
= |
{Corfu, Greece}, |
url |
= |
{http://hal.inria.fr/inria-00618222/fr/}, |
keyword |
= |
{Segmentation, Classification, skeletonization , pattern recognition, shape analysis} |
} |
Abstract :
High resolution satellite images provided by the last generation
sensors significantly increased the potential of almost
all the image information mining (IIM) applications related
to earth observation. This is especially true for the extraction
of road information, task of primary interest for many remote
sensing applications, which scope is more and more extended
to complex urban scenarios thanks to the availability of highly
detailed images. This context is particularly challenging due
to such factors as the variability of road visual appearence
and the occlusions from entities like trees, cars and shadows.
On the other hand, the peculiar geometry and morphology of
man-made structures, particularly relevant in urban areas, is
enhanced in high resolution images, making this kind of information
especially useful for road detection.
In this work, we provide a new insight on the use of morphological
image analysis for road extraction in complex urban
scenarios, and propose a technique for road segmentation
that only relies on this domain. The keypoint of the technique
is the use of skeletons as powerful descriptors for road objects:
the proposed method is based on an ad-hoc skeletonization
procedure that enhances the linear structure of road segments,
and extracts road objects by first detecting their skeletons
and then associating each of them with a region of the
image. Experimental results are presented on two different
high resolution satellite images of urban areas. |
|
2 - Brain tumor vascular network segmentation from micro-tomography. X. Descombes et F. Plouraboue et El Boustani Habdelhkim et Fonta Caroline et |LeDuc Geraldine et Serduc Raphael et Weitkamp Timm. Dans Internation Symposium of Biomedical Imaging (ISBI), Chicago, USA, avril 2011. Mots-clés : Segmentation, Markov random field, Tomography, Brain, vascular network. Copyright : IEEE
@INPROCEEDINGS{isbi11,
|
author |
= |
{Descombes, X. and Plouraboue, F. and Boustani Habdelhkim, El and Caroline, Fonta and Geraldine, |LeDuc and Raphael, Serduc and Timm, Weitkamp}, |
title |
= |
{Brain tumor vascular network segmentation from micro-tomography}, |
year |
= |
{2011}, |
month |
= |
{avril}, |
booktitle |
= |
{Internation Symposium of Biomedical Imaging (ISBI)}, |
address |
= |
{Chicago, USA}, |
url |
= |
{http://dx.doi.org/10.1109/ISBI.2011.5872596}, |
keyword |
= |
{Segmentation, Markov random field, Tomography, Brain, vascular network} |
} |
Abstract :
Micro-tomography produces high resolution images of biological structures such as vascular networks. In this paper, we present a new approach for segmenting vascular network into pathological and normal regions from considering their micro-vessel 3D structure only. We define and use a conditional random field for segmenting the output of a watershed algorithm. The tumoral and normal classes are thus characterized by their respective distribution of watershed region size interpreted as local vascular territories. |
|
3 - A markov random field model for extracting near-circular shapes. T. Blaskovics et Z. Kato et I. H. Jermyn. Dans Proc. IEEE International Conference on Image Processing (ICIP), Cairo, Egypt, novembre 2009. Mots-clés : Segmentation, Markov Random Fields, Shape prior.
@INPROCEEDINGS{Blaskovics09,
|
author |
= |
{Blaskovics, T. and Kato, Z. and Jermyn, I. H.}, |
title |
= |
{A markov random field model for extracting near-circular shapes}, |
year |
= |
{2009}, |
month |
= |
{novembre}, |
booktitle |
= |
{Proc. IEEE International Conference on Image Processing (ICIP)}, |
address |
= |
{Cairo, Egypt}, |
pdf |
= |
{http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=5413472}, |
keyword |
= |
{Segmentation, Markov Random Fields, Shape prior} |
} |
|
4 - Unsupervised One-Class SVM Using a Watershed Algorithm and Hysteresis Thresholding to Detect Burnt Areas. O. Zammit et X. Descombes et J. Zerubia. Dans Proc. International Conference on Pattern Recognition and Image Analysis (PRIA), Nizhny Novgorod, Russia, septembre 2008. Mots-clés : Classification, Segmentation, Support Vector Machines, Zones brûlées, Feux de foret, Imagerie satellitaire. Copyright :
@INPROCEEDINGS{zammit_pria_08,
|
author |
= |
{Zammit, O. and Descombes, X. and Zerubia, J.}, |
title |
= |
{Unsupervised One-Class SVM Using a Watershed Algorithm and Hysteresis Thresholding to Detect Burnt Areas}, |
year |
= |
{2008}, |
month |
= |
{septembre}, |
booktitle |
= |
{Proc. International Conference on Pattern Recognition and Image Analysis (PRIA)}, |
address |
= |
{Nizhny Novgorod, Russia}, |
pdf |
= |
{http://hal.inria.fr/inria-00316297/fr/}, |
keyword |
= |
{Classification, Segmentation, Support Vector Machines, Zones brûlées, Feux de foret, Imagerie satellitaire} |
} |
|
5 - Unsupervised Hierarchical Image Segmentation based on the TS-MRF model and Fast Mean-Shift Clustering. R. Gaetano et G. Scarpa et G. Poggi et J. Zerubia. Dans Proc. European Signal Processing Conference (EUSIPCO), Lausanne, Switzerland, août 2008. Mots-clés : Segmentation, Markov Random Fields, Mean Shift, Land Classification.
@INPROCEEDINGS{Gaetano2008,
|
author |
= |
{Gaetano, R. and Scarpa, G. and Poggi, G. and Zerubia, J.}, |
title |
= |
{Unsupervised Hierarchical Image Segmentation based on the TS-MRF model and Fast Mean-Shift Clustering}, |
year |
= |
{2008}, |
month |
= |
{août}, |
booktitle |
= |
{Proc. European Signal Processing Conference (EUSIPCO)}, |
address |
= |
{Lausanne, Switzerland}, |
pdf |
= |
{http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=7080521}, |
keyword |
= |
{Segmentation, Markov Random Fields, Mean Shift, Land Classification} |
} |
Abstract :
Tree-Structured Markov Random Field (TS-MRF) models have been recently proposed to provide a hierarchical multiscale description of images. Based on such a model, the unsupervised image segmentation is carried out by means of a sequence of nested class splits, where each class is modeled as a local binary MRF.
We propose here a new TS-MRF unsupervised segmentation technique which improves upon the original algorithm by selecting a better tree structure and eliminating spurious classes. Such results are obtained by using the Mean-Shift procedure to estimate the number of pdf modes at each node (thus allowing for a non-binary tree), and to obtain a more reliable initial clustering for subsequent MRF optimization. To this end, we devise a new reliable and fast clustering algorithm based on the Mean-Shift technique. Experimental results prove the potential of the proposed method. |
|
6 - Indexing of mid-resolution satellite images with structural attributes. A. Bhattacharya et M. Roux et H. Maitre et I. H. Jermyn et X. Descombes et J. Zerubia. Dans The International Society for Photogrammetry and Remote Sensing, Beijing, China, juillet 2008. Mots-clés : Landscape, Segmentation, Features, Extraction, Classification, Modelling.
@INPROCEEDINGS{Bhattacharya08,
|
author |
= |
{Bhattacharya, A. and Roux, M. and Maitre, H. and Jermyn, I. H. and Descombes, X. and Zerubia, J.}, |
title |
= |
{Indexing of mid-resolution satellite images with structural attributes}, |
year |
= |
{2008}, |
month |
= |
{juillet}, |
booktitle |
= |
{The International Society for Photogrammetry and Remote Sensing}, |
address |
= |
{Beijing, China}, |
pdf |
= |
{http://www-sop.inria.fr/members/Ian.Jermyn/publications/Bhattacharya08isprs.pdf}, |
keyword |
= |
{Landscape, Segmentation, Features, Extraction, Classification, Modelling} |
} |
Abstract :
Indexing and retrieval of satellite images relies on the extraction of appropriate information from the data about the entity of interest
(e.g. land cover type) and on the robustness of this extraction to nuisance variables. Entities in an image may be strongly correlated
with each other and can therefore be used to characterize geographical environments on the Earth’s surface.
The properties of road networks vary considerably from one geographical environment to another. The networks pertaining in a
satellite image can therefore be used to classify and retrieve such environments. In the work presented in this paper we have defined
7 such classes. These classes can be categorized as follows: 2 urban classes consisting of “Urban USA” and “Urban Europe”; 3
rural classes consisting of “Villages”, “Mountains” and “Fields”; an “Airports” class and a “Common” class (this can be considered
as a rejection class). These classes were then classified with the aid of geometrical and topological features computed from the road
networks occurring in them. In our work we have used two extraction methods simultaneously on an image to extract the road networks
pertaining in it. A set of 16 network features were computed from one extraction method and were categorized into 6 groups as follows:
6 measures of ‘density’, 4 measures of ‘curviness’, 2 measures of ‘homogeneity’, 1 measure of ‘length’, 2 measures of ‘distribution’
and 1 measure of ‘entropy’.
Due to certain limitations of these extraction methods there was a relative failure of network extraction in certain urban regions con-
taining narrow and dense road structures. This loss of information was circumvented by segmenting the urban regions and computing
a second set of geometrical and topological features from them. A set of 4 urban region features were computed and were categorized
into 3 groups as follows: 2 measures of ‘density’, 1 measure of ‘labels’ and 1 measure of ‘compactness’.
The 500 images (each of size 512x512 pixels) forming our database were selected from SPOT5 scenes with 5m resolution. From each
image a set of geometrical and topological features were computed from the road networks and urban regions. These features were
then used to classify the pre-defined geographical classes. Feature selection was done to avoid the burden of feature dimensionality
and increase the classification performance. A set of 20 features was selected from 36 features by Fisher Linear Discriminant (FLD)
analysis which gave the least classification error with an one-vs-rest linear Support Vector Machine (SVM).
The impact of spatial resolution and size of images on the feature set have been explored in this work. We took a closer look at the effect
of spatial resolution and size of images on the discriminative power of the feature set to classify the images belonging to the pre-defined
geographical classes. Tests were performed with feature selection by FLD and one-vs-rest linear SVM classification on a database with
images of 10m resolution. Another test was performed with feature selection by FLD and one-vs-rest linear SVM classification on a
database with 5m resolution images (each of size 256x256 pixels).
With the above mentioned approaches, we developed a novel method to classify large satellite images acquired by SPOT5 satellite (5m
resolution) with patches of images each of size 512x512 pixels extracted from them. There has been a large amount of work dedicated
to the classification of large satellite images at pixel level rather than considering image patches of different sizes. Classification of
image patches of different sizes from a large satellite image is a novel idea in the sense that the patches considered contain significant
coverage of a particular type of geographical environment.
Road networks and urban region features were computed from these image patches extracted from the large image. A one-vs-rest
Gaussian kernel SVM classification method was used to classify this large image. The classification results show that the image
patches were labeled with the class having the maximum geographical coverage of the area associated in the large image. The large
image was mapped into a “region matrix”, where each element of the matrix corresponds to a geographical class. This is a ‘hard’
classification and no inference can be drawn about the classification confidence.
In certain cases, this produces some anomalies, as a single patch may contain two or more different geographical coverages. In order
to have an estimate of these partial coverages, the output of the SVM was mapped into probabilities. These probability measures were
then studied to have a closer look at the classification accuracies. The results confirm that our method is able to classify a large image
into various geographical classes with a mean error of less than 10%.
Future studies can use operators to detect not only man-made structures like roads and urban areas, but also natural entities like rivers,
forests, etc. In this work we have restricted ourselves to a single resolution, but our methodology can be adapted to consider images
of higher resolutions from QuickBird and the future Pleiade satellite. At a better resolution it may be possible to extract different
structures like buildings, gardens, cross-roads, etc. This in turn will allow us to incorporate more classes to appropriately classify any
geographical environment. At an image resolution of 1m, we may imagine to have sub-classes of an existing class, e.g., classes like
urban Europe and urban USA can de divided into downtown, residential and industrial classes. |
|
7 - Extraction of main and secondary roads in VHR images using a higher-order phase field model. T. Peng et I. H. Jermyn et V. Prinet et J. Zerubia. Dans Proc. XXI ISPRS Congress, Part A, pages 215-22, Beijing, China, juillet 2008. Mots-clés : Reseaux routiers, Zones urbaines, Imagerie satellitaire, Segmentation, Modelling, Methodes variationnelles. Copyright : ISPRS
@INPROCEEDINGS{Peng08a,
|
author |
= |
{Peng, T. and Jermyn, I. H. and Prinet, V. and Zerubia, J.}, |
title |
= |
{Extraction of main and secondary roads in VHR images using a higher-order phase field model}, |
year |
= |
{2008}, |
month |
= |
{juillet}, |
booktitle |
= |
{Proc. XXI ISPRS Congress, Part A}, |
pages |
= |
{215-22}, |
address |
= |
{Beijing, China}, |
pdf |
= |
{http://www.isprs.org/proceedings/XXXVII/congress/3_pdf/33.pdf}, |
keyword |
= |
{Reseaux routiers, Zones urbaines, Imagerie satellitaire, Segmentation, Modelling, Methodes variationnelles} |
} |
Abstract :
This paper addresses the issue of extracting main and secondary road networks in dense urban areas from very high resolution (VHR, ~0.61m) satellite images. The difficulty with secondary roads lies in the low discriminative power of the grey-level distributions of road regions and the background, and the greater effect of occlusions and other noise on narrower roads. To tackle this problem, we use a previously developed higher-order active contour (HOAC) phase field model and augment it with an additional non-linear nonlocal term. The additional term allows separate control of road width and road curvature; thus more precise prior knowledge can be incorporated, and better road prolongation can be achieved for the same width. Promising results on QuickBird panchromatic images at reduced resolutions and comparisons with other models demonstrate the role and the efficiency of our new model. |
|
8 - Higher Order Active Contours and their Application to the Detection of Line Networks in Satellite Imagery. M. Rochery et I. H. Jermyn et J. Zerubia. Dans Proc. IEEE Workshop Variational, Geometric and Level Set Methods in Computer Vision, at ICCV, Nice, France, octobre 2003. Mots-clés : Ordre superieur, Contour actif, Forme, Reseaux routiers, Segmentation, A priori.
@INPROCEEDINGS{Rochery03a,
|
author |
= |
{Rochery, M. and Jermyn, I. H. and Zerubia, J.}, |
title |
= |
{Higher Order Active Contours and their Application to the Detection of Line Networks in Satellite Imagery}, |
year |
= |
{2003}, |
month |
= |
{octobre}, |
booktitle |
= |
{Proc. IEEE Workshop Variational, Geometric and Level Set Methods in Computer Vision}, |
address |
= |
{at ICCV, Nice, France}, |
pdf |
= |
{ftp://ftp-sop.inria.fr/ariana/Articles/rochery_vlsm03.pdf}, |
keyword |
= |
{Ordre superieur, Contour actif, Forme, Reseaux routiers, Segmentation, A priori} |
} |
Abstract :
We present a novel method for the incorporation of shape information
into active contour models, and apply it to the extraction
of line networks (e.g. road, water) from satellite imagery.
The method is based on a new class of contour energies.
These energies are quadratic on the space of one-chains
in the image, as opposed to classical energies, which are linear.
They can be expressed as double integrals on the contour,
and thus incorporate non-trivial interactions between
different contour points. The new energies describe families
of contours that share complex geometric properties, without
making reference to any particular shape. Networks fall
into such a family, and to model them we make a particular
choice of quadratic energy whose minima are reticulated.
To optimize the energies, we use a level set approach. The
forces derived from the new energies are non-local however,
thus necessitating an extension of standard level set methods.
Promising experimental results are obtained using real
images. |
|
9 - Unsupervised Image Segmentation via Markov Trees and Complex Wavelets. C. Shaffrey et N. Kingsbury et I. H. Jermyn. Dans Proc. IEEE International Conference on Image Processing (ICIP), Rochester, USA, septembre 2002. Mots-clés : Segmentation, Hidden Markov Model, Texture, Couleur.
@INPROCEEDINGS{ijking,
|
author |
= |
{Shaffrey, C. and Kingsbury, N. and Jermyn, I. H.}, |
title |
= |
{Unsupervised Image Segmentation via Markov Trees and Complex Wavelets}, |
year |
= |
{2002}, |
month |
= |
{septembre}, |
booktitle |
= |
{Proc. IEEE International Conference on Image Processing (ICIP)}, |
address |
= |
{Rochester, USA}, |
pdf |
= |
{http://www-sop.inria.fr/members/Ian.Jermyn/publications/Shaffrey02icip.pdf}, |
keyword |
= |
{Segmentation, Hidden Markov Model, Texture, Couleur} |
} |
Abstract :
The goal in image segmentation is to label pixels in an image based
on the properties of each pixel and its surrounding region. Recently
Content-Based Image Retrieval (CBIR) has emerged as an
application area in which retrieval is attempted by trying to gain
unsupervised access to the image semantics directly rather than
via manual annotation. To this end, we present an unsupervised
segmentation technique in which colour and texture models are
learned from the image prior to segmentation, and whose output
(including the models) may subsequently be used as a content
descriptor in a CBIR system. These models are obtained in a
multiresolution setting in which Hidden Markov Trees (HMT) are
used to model the key statistical properties exhibited by complex
wavelet and scaling function coefficients. The unsupervised Mean
Shift Iteration (MSI) procedure is used to determine a number of
image regions which are then used to train the models for each
segmentation class. |
|
10 - Globally optimal regions and boundaries. I. H. Jermyn et H. Ishikawa. Dans Proc. IEEE International Conference on Computer Vision (ICCV), 1999. Mots-clés : global, optimum, Graphe, Cycle, Ratio, Segmentation. Copyright :
@INPROCEEDINGS{Jermyn99iccv,
|
author |
= |
{Jermyn, I. H. and Ishikawa, H.}, |
title |
= |
{Globally optimal regions and boundaries}, |
year |
= |
{1999}, |
booktitle |
= |
{Proc. IEEE International Conference on Computer Vision (ICCV)}, |
pdf |
= |
{http://www-sop.inria.fr/members/Ian.Jermyn/publications/Jermyn99iccv.pdf}, |
keyword |
= |
{global, optimum, Graphe, Cycle, Ratio, Segmentation} |
} |
Abstract :
We propose a new form of energy functional for the segmentation
of regions in images, and an efficient method for
finding its global optima. The energy can have contributions
from both the region and its boundary, thus combining
the best features of region- and boundary-based approaches
to segmentation. By transforming the region energy
into a boundary energy, we can treat both contributions
on an equal footing, and solve the global optimization
problem as a minimum mean weight cycle problem on
a directed graph. The simple, polynomial-time algorithm
requires no initialization and is highly parallelizable. |
|
haut de la page
7 Rapports de recherche et Rapports techniques |
1 - Hierarchical finite-state modeling for texture segmentation with application to forest classification. G. Scarpa et M. Haindl et J. Zerubia. Research Report 6066, INRIA, INRIA, France, décembre 2006. Mots-clés : Texture, Segmentation, Co-occurrence matrix, Approche structurelle, MCMC, Synthesis.
@TECHREPORT{scarparr06,
|
author |
= |
{Scarpa, G. and Haindl, M. and Zerubia, J.}, |
title |
= |
{Hierarchical finite-state modeling for texture segmentation with application to forest classification}, |
year |
= |
{2006}, |
month |
= |
{décembre}, |
institution |
= |
{INRIA}, |
type |
= |
{Research Report}, |
number |
= |
{6066}, |
address |
= |
{INRIA, France}, |
url |
= |
{https://hal.inria.fr/inria-00118420}, |
keyword |
= |
{Texture, Segmentation, Co-occurrence matrix, Approche structurelle, MCMC, Synthesis} |
} |
Abstract :
In this research report we present a new model for texture representation which is particularly well suited for image analysis and segmentation. Any image is first discretized and then a hierarchical finite-state region-based model is automatically coupled with the data by means of a sequential optimization scheme, namely the Texture Fragmentation and Reconstruction (TFR) algorithm. The TFR algorithm allows to model both intra- and inter-texture interactions, and eventually addresses the segmentation task in a completely unsupervised manner. Moreover, it provides a hierarchical output, as the user may decide the scale at which the segmentation has to be given. Tests were carried out on both natural texture mosaics provided by the Prague Texture Segmentation Datagenerator Benchmark and remote-sensing data of forest areas provided by the French National Forest Inventory (IFN). |
|
2 - Detecting Codimension-two Objects in an Image with Ginzburg-Landau Models. G. Aubert et J.F. Aujol et L. Blanc-Féraud. Rapport de Recherche 5254, INRIA, France, juillet 2004. Mots-clés : Modele de Ginzburg-Landau, Images biologiques, Segmentation, Equation aux derivees partielles.
@TECHREPORT{5254,
|
author |
= |
{Aubert, G. and Aujol, J.F. and Blanc-Féraud, L.}, |
title |
= |
{Detecting Codimension-two Objects in an Image with Ginzburg-Landau Models}, |
year |
= |
{2004}, |
month |
= |
{juillet}, |
institution |
= |
{INRIA}, |
type |
= |
{Research Report}, |
number |
= |
{5254}, |
address |
= |
{France}, |
url |
= |
{https://hal.inria.fr/inria-00070744}, |
pdf |
= |
{https://hal.inria.fr/file/index/docid/70744/filename/RR-5254.pdf}, |
ps |
= |
{https://hal.inria.fr/docs/00/07/07/44/PS/RR-5254.ps}, |
keyword |
= |
{Modele de Ginzburg-Landau, Images biologiques, Segmentation, Equation aux derivees partielles} |
} |
Résumé :
Dans cet article, nous proposons a nouveau modèle mathématique pour détecter dans une image les singularités de codimension supérieure ou égale à deux. Cela signifie que nous voulons détecter des points dans des images 2-D, ou des points et des courbes dans des images 3-D. Nous nous inspirons des modèles de Ginzburg-Landau (GL). Ces derniers se sont révélés efficace pour modéliser de nombreux phénomènes physiques. Nous introduisons le modèle, nous énonçons ses propriétés mathématiques, et nous donnons des résultats expérimentaux illustrant les performances du modèle. |
Abstract :
In this paper, we propose a new mathematical model for detecting in an image singularities of codimension greater than or equal to two. This means we want to detect points in a 2-D image or points and curves in a 3-D image. We drew one's inspiration from Ginzburg-Landau (G-L) models which have proved their efficiency for modeling many phenomena in physics. We introduce the model, state its mathematical properties and give some experimental results demonstrating its capability. |
|
3 - A Probabilistic Framework for Adaptive Texture Description. K. Brady et I. H. Jermyn et J. Zerubia. Rapport de Recherche 4920, INRIA, France, septembre 2003. Mots-clés : Segmentation, Texture, Paquet d'ondelettes.
@TECHREPORT{4920,
|
author |
= |
{Brady, K. and Jermyn, I. H. and Zerubia, J.}, |
title |
= |
{A Probabilistic Framework for Adaptive Texture Description}, |
year |
= |
{2003}, |
month |
= |
{septembre}, |
institution |
= |
{INRIA}, |
type |
= |
{Research Report}, |
number |
= |
{4920}, |
address |
= |
{France}, |
url |
= |
{https://hal.inria.fr/inria-00071659}, |
pdf |
= |
{https://hal.inria.fr/file/index/docid/71659/filename/RR-4920.pdf}, |
ps |
= |
{https://hal.inria.fr/docs/00/07/16/59/PS/RR-4920.ps}, |
keyword |
= |
{Segmentation, Texture, Paquet d'ondelettes} |
} |
Résumé :
Ce rapport présente le développement d'un nouveau cadre probabiliste cohérent pour la description adaptative de texture. En partant d'une distribution de probabilité sur un espace d'images infinies, nous générons une distribution sur des régions finies par marginalisation. Pour une distribution gaussienne, les contraintes de calcul imposées par la diagonalisation nous conduisent naturellement à des modèles utilisant des paquets d'ondelettes adaptatifs. Ces modèles reflètent les principales périodicités présentes dans les textures et permettent également d'avoir des corrélations à longue portée tout en préservant l'indépendance des coefficients des paquets d'ondelettes. Nous avons appliqué notre méthode à la segmentation. Deux types de données figurent dans notre ensemble de test: des mosaïques synthétiques de Brodatz et des images satellitaires haute résolution. Dans le cas des textures synthétiques, nous utilisons la version non-décimée de la transformée en paquets d'ondelettes afin de diagonaliser la distribution gaussienne de manière efficace, bien qu'approximative. Cela nous permet d'effectuer une classification de la mosaique pixel par pixel. Une étape de régularisation est ensuite effectuée afin d'arriver à un résultat de segmentation final plus lisse. Afin d'obtenir les meilleurs résultats possibles dans le cas de données réelles, la moyenne de la distribution est ensuite introduite dans le modèle. L'approximation faite pour la classification des mosaiques de textures synthetiques a été testée sur des images réelles, mais les résultats obtenus n'étaient pas satisfaisants. C'est pourquoi nous avons introduit, pour ce type de données, une technique de classification heuristique basée sur la transformée en paquets d'ondelettes décimée. Les résultats de segmentation sont ensuite régularisés à l'aide de la même méthode que dans le cas synthétique. Nous présentons les résultats pour chaque type de données et concluons par une discussion. |
Abstract :
This report details the development of a probabilistic framework for adaptive texture description. Starting with a probability distribution on the space of infinite images, we generate a distribution on finite regions by marginalisation. For a Gaussian distribution, the computational requirement of diagonalisation leads naturally to adaptive wavelet packet models which capture the principal periodicities present in the textures and allow long-range correlations while preserving the independence of the wavelet packet coefficients. These models are then applied to the task of segmentation. Two data types are included in our test bed: synthetic Brodatz mosaics and high-resolution satellite images. For the case of the synthetic textures, undecimated versions of the wavelet packet transform are used to diagonalise the Gaussian distribution efficiently, albeit approximately. This enables us to perform a pixelwise classification of the mosaics. A regularisation step is then implemented in order to arrive at a smooth final segmentation. In order to obtain the best possible results for the real dataset, the mean of the distribution is included in the model. The approximation made for the classification of the synthetic texture mosaics is tested on the remote sensing images, but it produces unsatisfactory results. Therefore we introduce a heuristic classification technique for this dataset, based on a decimated wavelet packet transform. The resulting segmentation is then regularised using the same method as in the synthetic case. Results are presented for both types of data and a discussion follows. |
|
4 - The Methodology and Practice of the Evaluation of Image Retrieval Systems and Segmentation Methods. I. H. Jermyn et C. Shaffrey et N. Kingsbury. Rapport de Recherche 4761, INRIA, France, mars 2003. Mots-clés : Base de donnees Image, Segmentation, Semantique.
@TECHREPORT{4761,
|
author |
= |
{Jermyn, I. H. and Shaffrey, C. and Kingsbury, N.}, |
title |
= |
{The Methodology and Practice of the Evaluation of Image Retrieval Systems and Segmentation Methods}, |
year |
= |
{2003}, |
month |
= |
{mars}, |
institution |
= |
{INRIA}, |
type |
= |
{Research Report}, |
number |
= |
{4761}, |
address |
= |
{France}, |
url |
= |
{https://hal.inria.fr/inria-00071825}, |
pdf |
= |
{https://hal.inria.fr/file/index/docid/71825/filename/RR-4761.pdf}, |
ps |
= |
{https://hal.inria.fr/docs/00/07/18/25/PS/RR-4761.ps}, |
keyword |
= |
{Base de donnees Image, Segmentation, Semantique} |
} |
Résumé :
La recherche d'images par le contenu est importante pour deux raisons. Premièrement, la croissance d'archives d'images fréquemment citée dans beaucoup d'applications, et l'expansion rapide du Web, signifient qu'il est nécessaire d'utiliser des systèmes de recherche efficaces pour les bases de données afin que la masse de données accumulée soit utile. Deuxièmement, la recherche dans les bases de données image pose des questions importantes liées à la vision par ordinateur : une recherche efficace demande une véritable compréhension des images. Pour ces raisons, l'évaluation des systèmes de recherche dans les bases de données image devient une priorité. Il existe déjà une littérature importante évaluant des systèmes spécifiques, mais peu de discussions sont publiées sur les méthodes d'évaluation en soi. Dans la première partie de ce rapport, nous proposons un cadre dans lequel ces sujets peuvent être abordés, nous analysons des méthodologies d'évaluation possibles, indiquant quand elles sont pertinentes et quand elles ne le sont pas, et nous critiquons la technique «query-by-example» et les méthodes d'évaluation qui s'y rapportent. Dans la deuxième partie du rapport, nous appliquons les résultats de cette analyse à une collection spécifique d'images. Cette collection est problématique mais typique: il n'existe pas de vérité terrain sémantique. Considérant la recherche fondée sur la segmentation d'image, nous présentons une nouvelle méthode pour son évaluation. Contrairement aux méthodes d'évaluation qui reposent sur l'existence ou la création d'une vérité terrain, la méthodologie proposée utilise des sujets humains pour un test psychovisuel qui compare les résultats des différentes méthodes de segmentation. Le test est con u pour répondre à deux questions : existe-t-il une segmentation «meilleure» que les autres et si oui qu'apprenons-nous des méthodes de segmentation pour la recherche dans des bases de données image? Les résultats confirment la cohérence des jugements humains, permettant ainsi une évaluation significative. |
Abstract :
Content-Based Image Retrieval is important for two reasons. First, the oft-cited growth of image archives in many fields, and the rapid expansion of the Web, mean that successful image retrieval systems are fast becoming a necessity if the mass of accumulated data is to be useful. Second, database retrieval provides a framework within which the important questions of machine vision are brought into focus: successful retrieval is likely to require genuine image understanding. In view of these points, the evaluatio- n of retrieval systems becomes a matter of priority. There is already a substantial literature evaluating specific systems, but little high-level discussion of the evaluation methodologies themselves seems to have taken place. In the first part of the report, we propose a framework within which such issues can be addressed, analyse possible evaluation methodologies, indicate where they are appropriate and where they are not, and critique query-by-example and evaluation methodologies related to it. In the second part of the report, we apply the results of this analysis to a particular dataset. The dataset is problematic but typical: no ground truth is available for its semantics. Considering retrieval based on image segmentation- s, we present a novel method for its evaluation. Unlike methods of evaluation that rely on the existence or creation of ground truth, the proposed evaluatio- n procedure subjects human subjects to a psychovisual test comparing the results of different segmentation schemes. The test is designed to answer two questions: does consensus about a `best' segmentation exist, and if it does, what do we learn about segmentation schemes for retrieval? The results confirm that human subjects are consistent in their judgements, thus allowing meaningful evaluation. |
|
5 - Simulation de processus objets : Etude de faisabilité pour une application à la segmentation d'image. M. Imberty et X. Descombes. Rapport de Recherche 3881, Inria, février 2000. Mots-clés : Processus ponctuels marques, Geometrie stochastique, Segmentation.
@TECHREPORT{xd00im,
|
author |
= |
{Imberty, M. and Descombes, X.}, |
title |
= |
{Simulation de processus objets : Etude de faisabilité pour une application à la segmentation d'image}, |
year |
= |
{2000}, |
month |
= |
{février}, |
institution |
= |
{Inria}, |
type |
= |
{Research Report}, |
number |
= |
{3881}, |
url |
= |
{https://hal.inria.fr/inria-00072772}, |
pdf |
= |
{https://hal.inria.fr/file/index/docid/72772/filename/RR-3881.pdf}, |
ps |
= |
{https://hal.inria.fr/docs/00/07/27/72/PS/RR-3881.ps}, |
keyword |
= |
{Processus ponctuels marques, Geometrie stochastique, Segmentation} |
} |
Résumé :
Dans cette étude, nous comparons l'efficacité de deux techniques de simulation par chaînes de Markov (MCMC) de processus aléatoires sur des ensembles d'objets géométriques : l'algorithme de naissance-mort et celui de Metropolis-- Hastings-Green. Les comparaisons sont effectuées sur différents modèles de processus objets de type attractif présentant un intérêt en traitement d'image. Nous appliquons ensuite ces méthodes de simulation à la segmentation d'image. Pour cela, nous nous plaçons dans le cadre bayésien : nous définisson- s donc un modèle a priori attractif simple sur des objets rectangulaires, ainsi qu'un terme d'attache aux données garantissant l'adéquation des objets à l'image. Nous utilisons ensuite un recuit simulé pour extraire les différentes zones de l'image. Des tests sont effectués sur des images synthétiques. |
Abstract :
In this study, we compare the efficiency of two algorithms using Monte Carlo Markov chains methods in order to simulate random processes of geometric- al objects sets : the algorithm of birth and death and the dynamics of Metropolis-Hastings-Green. The comparisons are carried out on various object models for clustered patterns, which could be of interest in image processing. Then we apply these methods of simulation to image segmentation, using the bayesian approach : thus we define a simple prior model on rectangul- ar objects, as well as a posterior probability guaranteeing the adequacy of the objects to the data. We finally use a stochastic annealing to extract the various zones of the image. Some tests are performed on synthetic data. |
|
6 - Isotropic Properties of Some Multi-body Interaction Models: Two Quality Criteria for Markov Priors in Image Processing. X. Descombes et E. Pechersky. Rapport de Recherche 3752, Inria, août 1999. Mots-clés : Champs de Gibbs, Segmentation, Restauration.
@TECHREPORT{xd99k,
|
author |
= |
{Descombes, X. and Pechersky, E.}, |
title |
= |
{Isotropic Properties of Some Multi-body Interaction Models: Two Quality Criteria for Markov Priors in Image Processing}, |
year |
= |
{1999}, |
month |
= |
{août}, |
institution |
= |
{Inria}, |
type |
= |
{Research Report}, |
number |
= |
{3752}, |
url |
= |
{http://hal.inria.fr/inria-00072910}, |
pdf |
= |
{http://hal.inria.fr/docs/00/07/29/10/PDF/RR-3752.pdf}, |
ps |
= |
{http://hal.inria.fr/docs/00/07/29/10/PS/RR-3752.ps}, |
keyword |
= |
{Champs de Gibbs, Segmentation, Restauration} |
} |
Résumé :
Les champs de Gibbs sont très utilisés en traitement d'image à la fois pour la segmentation et la restauration. Définis sur la trâme discrète sous-jacente à l'image, ils présentent un comportement non isotrope. Dans ce rapport, nous étudions et quantifions cette non-isotropie, pour des modèles avec des interactions 3x3, en calculant la tension de bord en fonction de l'angle d'une droite séparant le plan en deux parties contenant une phase différente. De cette étude, nous dérivons deux critères quantitatifs d'anisotropie des modèles. Nous calculons ensuite la forme d'une goutte d'une phase immergée dans une autre phase à la température nulle pour les différents modèles, et étudions la non isotropie des formes obtenues. Pour finir, les artéfacts induits par cette non-isotropie sont mis en évidence sur des exemples de segmentation et de restauration d'image. |
Abstract :
Gibbs Fields are widely used in image processing for both segmentation and restoration. Defined on a discrete lattice representing the image they exhibit a non-isotropic behavior. Herein, we study and quantify this non-isotropy by computing the boundary tension as a function of the angle of a line separating the plane in two parts containing a different phase. From this study, we derive two quantitative criteria of the non isotropy of the model. We then compute the shape at zero temperature of a droplet of one phase within the other phase and study the non-isotropy of the shape for the different models. Finally, we show the artifacts due to this non-isotropic behavior for image segmentation and restoration. |
|
7 - Indexing and retrieval in multimedia libraries through parametric texture modeling using the 2D Wold decomposition. R. Stoica et J. Zerubia et J.M. Francos. Rapport de Recherche 3594, Inria, décembre 1998. Mots-clés : Champs de Markov, Texture, Segmentation, Indexation.
@TECHREPORT{stoica98,
|
author |
= |
{Stoica, R. and Zerubia, J. and Francos, J.M.}, |
title |
= |
{Indexing and retrieval in multimedia libraries through parametric texture modeling using the 2D Wold decomposition}, |
year |
= |
{1998}, |
month |
= |
{décembre}, |
institution |
= |
{Inria}, |
type |
= |
{Research Report}, |
number |
= |
{3594}, |
url |
= |
{https://hal.inria.fr/inria-00073085}, |
pdf |
= |
{https://hal.inria.fr/file/index/docid/73085/filename/RR-3594.pdf}, |
ps |
= |
{https://hal.inria.fr/docs/00/07/30/85/PS/RR-3594.ps}, |
keyword |
= |
{Champs de Markov, Texture, Segmentation, Indexation} |
} |
Résumé :
Ce rapport présente une méthode paramétrique permettant de faire de l'indexati- on et de la recherche dans une base de données multimédia. L'indexation (étiquetage) et la recherche de données multimédia sont réalisées grâce à la modélisation paramétrique de textures qui se trouvent dans les images de la base de données. Les textures sont caracterisées par des paramètres qui servent d'indices pour la recherche dans la base de données. Afin de pouvoir identifier les différentes régions texturées d'une image et estimer les paramètres correspondants, un algorithme de segmentation-estimatio- n est proposé dans ce rapport, qui fait appel à une décomposition de Wold 2D pour le modèle de texture et à un modèle markovien pour l'étiquetage. L'indexation nécessite de définir une distance entre les images. Une nouvelle distance, inspirée de la distance de Kullback, est décrite dans ce rapport. Elle utilise les paramètres estimés correspondants au modèle 2D de chaque texture. Les résultats obtenus relativement à la segmentation et à l'indexatio- n sont proches de ceux obtenus par un opérateur humain. |
Abstract :
This paper presents a parametric method for indexing and retrieval of multimedia data in digital libraries. %Indexing (labeling) and retrieval %of multimedia data, based on the properties %of the imagery components of the stored data record, are derived. Indexing (labeling) and retrieval of the multimedia data are performed using parametric modeling of the textured segments found in the data imagery components. The estimated parametric models of the textured segments serve as their indices, and hence as indices of the entire image, as well as of the multimedia record which the image is part thereof. To achieve the ability to identify textured image regions and estimate their parameters, a joint segmentation-estimation algorithm that combines the 2-D Wold decomposition based texture model with a Markovian labeling process, is derived. Ordering and indexing of images require a definition of a distance measure between images. Using the framework of the Kullback distance between probability distributions, a new rigorous distance measure between textures is derived. The distance between any two textured image segments is evaluated using their estimated parametric models. The proposed segmentation, distance evaluation, and indexing methods are shown to produce comparable results to those obtained by a human viewer. |
|
haut de la page
Ces pages sont générées par
|