Publications sur Extraction d'objets

Résultat de la recherche dans la liste des publications :

• 1 Article
• 3 Thèses de Doctorat et Habilitations
• 10 Articles de conférence
• 9 Rapports de recherche et Rapports techniques
• 2 Articles de collection ou Chapitres de livres

nouvelle recherche | bibtex

navigation : 1 - 2 - 3 affichage : 10 par page | 20 par page | 50 par page

Article

1 - Object Extraction Using a Stochastic Birth-and-Death Dynamics in Continuum. X. Descombes et R. Minlos et E. Zhizhina. Journal of Mathematical Imaging and Vision, 33(3): pages 347-359, 2009. Mots-clés : birth and death process, Processus ponctuels marques, Extraction d'objets. Copyright : Springer @ARTICLE{DZM08, author = {Descombes, X. and Minlos, R. and Zhizhina, E.}, title = {Object Extraction Using a Stochastic Birth-and-Death Dynamics in Continuum}, year = {2009}, journal = {Journal of Mathematical Imaging and Vision}, volume = {33}, number = {3}, pages = {347-359}, pdf = {http://dx.doi.org/10.1007/s10851-008-0117-y}, keyword = {birth and death process, Processus ponctuels marques, Extraction d'objets} } Abstract : We define a new birth and death dynamics dealing with configurations of disks in the plane. We prove the convergence of the continuous process and propose a discrete scheme converging to the continuous case. This framework is developed to address image processing problems consisting in detecting a configuration of objects from a digital image. The derived algorithm is applied for tree crown extraction and bird detection from aerial images. The performance of this approach is shown on real data.

haut de la page

3 Thèses de Doctorat et Habilitations

1 - Etude du couvert forestier par processus ponctuels marqués. G. Perrin. Thèse de Doctorat, Ecole Centrale Paris, octobre 2006. Mots-clés : Extraction de Houppiers, Processus ponctuels marques, Geometrie stochastique, Extraction d'objets, RJMCMC. @PHDTHESIS{perrin_phd06, author = {Perrin, G.}, title = {Etude du couvert forestier par processus ponctuels marqués}, year = {2006}, month = {octobre}, school = {Ecole Centrale Paris}, url = {http://www-sop.inria.fr/ariana/personnel/Guillaume.Perrin/resume.php}, pdf = {http://www-sop.inria.fr/ariana/personnel/Guillaume.Perrin/DOWNLOADS/these_perrin_2006.pdf}, keyword = {Extraction de Houppiers, Processus ponctuels marques, Geometrie stochastique, Extraction d'objets, RJMCMC} } Résumé : Cette thèse aborde le problème de l'extraction d'arbres à partir d'images aériennes InfraRouge Couleur (IRC) de forêts. Nos modèles reposent sur l'utilisation de processus objets ou processus ponctuels marqués. Il s'agit de variables aléatoires dont les réalisations sont des configurations d'objets géométriques. Une fois l'objet géométrique de référence choisi, nous définissons l'énergie du processus par le biais d'un terme a priori, modélisant les contraintes sur les objets et leurs interactions, ainsi qu'un terme image. Nous échantillonnons le processus objet grâce à un algorithme de type Monte Carlo par Chaînes de Markov à sauts réversibles (RJMCMC), optimisé par un recuit simulé afin d'extraire la meilleure configuration d'objets, qui nous donne l'extraction recherchée. Dans ce manuscrit, nous proposons différents modèles d'extraction de houppiers, qui extraient des informations à l'échelle de l'arbre selon la densité du peuplement. Dans les peuplements denses, nous présentons un processus d'ellipses, et dans les zones de plus faible densité, un processus d'ellipsoïdes. Nous obtenons ainsi le nombre d'arbres, leur localisation, le diamètre de la couronne et leur hauteur pour les zones non denses. Les algorithmes automatiques résultant de cette modélisation sont testés sur des images IRC très haute résolution fournies par l'Inventaire Forestier National (IFN). Abstract : This thesis addresses the problem of tree crown extraction from Colour InfraRed (CIR) aerial images of forests. Our models are based on object processes, otherwise known as marked point processes. These mathematical objects are random variables whose realizations are configurations of geometrical shapes. This approach yields an energy minimization problem, where the energy is composed of a regularization term (prior density), which introduces some constraints on the objects and their interactions, and a data term, which links the objects to the features to be extracted. Once the reference object has been chosen, we sample the process and extract the best configuration of objects with respect to the energy, using a Reversible Jump Markov Chain Monte Carlo (RJMCMC) algorithm embedded in a Simulated Annealing scheme. We propose different models for tree crown extraction depending on the density of the stand. In dense areas, we use an ellipse process, while in sparse vegetation an ellipsoïd process is used. As a result we obtain the number of stems, their position, the diameters of the crowns and the heights of the trees for sparse areas. The resulting algorithms are tested on high resolution CIR aerial images provided by the French National Forest Inventory (IFN).

2 - Processus Ponctuels Marqués pour l'Extraction Automatique de Caricatures de Bâtiments à partir de Modèles Numériques d'Elévation. M. Ortner. Thèse de Doctorat, Universite de Nice Sophia Antipolis, octobre 2004. Mots-clés : Processus ponctuels marques, Extraction d'objets, Batiments, Modele numerique d'elevation (MNE), RJMCMC, Geometrie stochastique. @PHDTHESIS{mortner_these, author = {Ortner, M.}, title = {Processus Ponctuels Marqués pour l'Extraction Automatique de Caricatures de Bâtiments à partir de Modèles Numériques d'Elévation}, year = {2004}, month = {octobre}, school = {Universite de Nice Sophia Antipolis}, url = {https://hal.inria.fr/tel-00189803}, pdf = {http://hal.inria.fr/docs/00/18/98/03/PDF/These_Ortner.pdf}, keyword = {Processus ponctuels marques, Extraction d'objets, Batiments, Modele numerique d'elevation (MNE), RJMCMC, Geometrie stochastique} } Résumé : Cette thèse se place dans un cadre de reconstruction urbaine et propose un corpus algorithmique pour extraire des formes simples sur les Modèles Numériques d'Elévation. Ce type de données décrit le relief d'une zone urbaine par une grille régulière de points à chacun desquels est associée une information de hauteur. Les modèles utilisés reposent sur l'utilisation de processus ponctuels marqués. Il s'agit de variables aléatoires dont les réalisations sont des configurations d'objets géométriques. Ces modèles permettent d'introduire des contraintes sur la forme des objets recherchés dans une image ainsi qu'un terme de régularisation modélisé par des interactions entre les objets. Une énergie peut être associée aux configurations d'objets et la configuration minimisant cette énergie trouvée au moyen d'un recuit-simulé couplé à un échantillonneur de type Monte Carlo par Chaîne de Markov à sauts réversibles (RJMCMC). Nous proposons quatre modèles pour extraire des caricatures de bâtiments à partir de descriptions altimétriques de zones urbaines denses. Chaque modèle est constitué par une forme d'objet, une énergie d'attache aux données et une énergie de régularisation. Les deux premiers modèles permettent d'extraire des formes simples (rectangles) en utilisant une contrainte d'homogénéité pour l'un et une détection des discontinuités pour l'autre. Le troisième modèle modélise les bâtiments par une forme polyhédrique. Le dernier modèle s'intéresse à l'apport d'une coopération entre des objets simples. Les algorithmes obtenus, automatiques, sont évalués sur des données réelles fournies par l'IGN (MNE Laser et optiques de différentes qualités). Abstract : The context of this thesis is the reconstruction of urban areas from images. It proposes a set of algorithms for extracting simple shapes from Digital Elevation Models (DEM). DEMs describe the altimetry of an urban area by a grid of points, each of which has a height associated to it. The proposed models are based on marked point processes. These mathematical objects are random variables whose realizations are configurations of geometrical shapes. Using these processes, we can introduce constraints on the shape of the objects to be detected in an image, and a regularizing term incorporating geometrical interactions between objects. An energy can be associated to each object configuration, and the global minima of this energy can then be found by applying simulated annealing to a Reversible Jump Monte Carlo Markov Chain sampler (RJMCMC). We propose four different models for extracting the outlines of buildings from altimetric descriptions of dense urban areas. Each of these models is constructed from an object shape, a data energy, and a regularizing energy. The first two models extract simple shapes (rectangles) using, respectively, a homogeneity constraint and discontinuity detection. The third model looks for three-dimensional polyhedral buildings. The last model uses cooperation between two types of objects, rectangles and segments. The resulting algorithms are evaluated on real data provided by the French National Geographic Institute (a laser DEM and optical DEMs of differing quality).

3 - Extraction de Réseaux Linéiques à partir d'Images Satellitaires et Aériennes par Processus Ponctuels Marqués. C. Lacoste. Thèse de Doctorat, Universite de Nice Sophia Antipolis, septembre 2004. Mots-clés : Geometrie stochastique, Extraction d'objets, RJMCMC, Reseaux lineiques, Recuit Simule, Processus ponctuels marques. @PHDTHESIS{lacoste_these, author = {Lacoste, C.}, title = {Extraction de Réseaux Linéiques à partir d'Images Satellitaires et Aériennes par Processus Ponctuels Marqués}, year = {2004}, month = {septembre}, school = {Universite de Nice Sophia Antipolis}, url = {https://hal.inria.fr/tel-00261397}, pdf = {http://hal.inria.fr/docs/00/26/13/97/PDF/THESE_CAROLINE_LACOSTE.pdf}, keyword = {Geometrie stochastique, Extraction d'objets, RJMCMC, Reseaux lineiques, Recuit Simule, Processus ponctuels marques} } Résumé : Cette thèse aborde le problème de l'extraction non supervisée des réseaux linéiques (routes, rivières, etc.) à partir d'images satellitaires et aériennes. Nous utilisons des processus objet, ou processus ponctuels marqués, comme modèles a priori. Ces modèles permettent de bénéficier de l'apport d'un cadre stochastique (robustesse au bruit, corpus algorithmique, etc.) tout en manipulant des contraintes géométriques fortes. Un recuit simulé sur un algorithme de type Monte Carlo par Chaîne de Markov (MCMC) permet une optimisation globale sur l'espace des configurations d'objets, indépendamment de l'initialisation. Nous proposons tout d'abord une modélisation du réseau linéique par un processus dont les objets sont des segments interagissant entre eux. Le modèle a priori est construit de façon à exploiter au mieux la topologie du réseau recherché au travers de potentiels fondés sur la qualité de chaque interaction. Les propriétés radiométriques sont prises en compte dans un terme d'attache aux données fondé sur des mesures statistiques. Nous étendons ensuite cette modélisation à des objets plus complexes. La manipulation de lignes brisées permet une extraction plus précise du réseau et améliore la détection des bifurcations. Enfin, nous proposons une modélisation hiérarchique des réseaux hydrographiques dans laquelle les affluents d'un fleuve sont modélisés par un processus de lignes brisées dans le voisinage de ce fleuve. Pour chacun des modèles, nous accélérons la convergence de l'algorithme MCMC par l'ajout de perturbations adaptées. La pertinence de cette modélisation par processus objet est vérifiée sur des images satellitaires et aériennes, optiques et radar. Abstract : This thesis addresses the problem of the unsupervised extraction of line networks (roads, rivers, etc.) from remotely sensed images. We use object processes, or marked point processes, as prior models. These models benefit from a stochastic framework (robustness w.r.t. noise, algorithms, etc.) while incorporating strong geometric constraints. Optimization is done via simulated annealing using a Reversible Jump Markov Chain Monte Carlo (RJMCMC) algorithm, without any specific initialization. We first propose to model line networks by a process whose objects are interacting line segments. The prior model is designed to exploit as fully as possible the topological properties of the network under consideration through potentials based on the quality of each interaction. The radiometric properties of the network are modeled using a data term based on statistical measures. We then extend this model to more complex objects. The use of broken lines improves the detection of network junctions and increases the accuracy of the extracted network. Finally, we propose a hierarchical model of hydrographic networks in which the tributaries of a given river are modeled by a process of broken lines in the neighborhood of this river. For each model, we accelerate convergence of the RJMCMC algorithm by using appropriate perturbations. We show experimental results on aerial and satellite images (optical and radar data) to verify the relevance of the object process models.

haut de la page

10 Articles de conférence

1 - Extraction of arbitrarily shaped objects using stochastic multiple birth-and-death dynamics and active contours. M. S. Kulikova et I. H. Jermyn et X. Descombes et E. Zhizhina et J. Zerubia. Dans Proc. IS&T/SPIE Electronic Imaging, San Jose, USA, janvier 2010. Mots-clés : Extraction d'objets, Processus ponctuels marques, Shape prior, Contour actif, birth-and-death dynamics. Copyright : Copyright 2010 by SPIE and IS&T. This paper was published in the proceedings of IS&T/SPIE Electronic Imaging 2010 Conference in San Jose, USA, and is made available as an electronic reprint with permission of SPIE and IS&T. One print or electronic copy may be made for personal use only. Systematic or multiple reproduction, distribution to multiple locations via electronic or other means, duplication of any material in this paper for a fee or for commercial purposes, or modification of the content of the paper are prohibited. @INPROCEEDINGS{Kulikova10a, author = {Kulikova, M. S. and Jermyn, I. H. and Descombes, X. and Zhizhina, E. and Zerubia, J.}, title = {Extraction of arbitrarily shaped objects using stochastic multiple birth-and-death dynamics and active contours}, year = {2010}, month = {janvier}, booktitle = {Proc. IS&T/SPIE Electronic Imaging}, address = {San Jose, USA}, pdf = {http://hal.archives-ouvertes.fr/docs/00/46/54/72/PDF/Kulikova_SPIE2010.pdf}, keyword = {Extraction d'objets, Processus ponctuels marques, Shape prior, Contour actif, birth-and-death dynamics} } Abstract : We extend the marked point process models that have been used for object extraction from images to arbitrarily shaped objects, without greatly increasing the computational complexity of sampling and estimation. From an alternative point of view, the approach can be viewed as an extension of the active contour methodology to an a priori unknown number of objects. Sampling and estimation are based on a stochastic birth-and-death process defined on the configuration space of an arbitrary number of objects, where the objects are defined by the image data and prior information. The performance of the approach is demonstrated via experimental results on synthetic and real data.

2 - A marked point process model with strong prior shape information for extraction of multiple, arbitrarily-shaped objects. M. S. Kulikova et I. H. Jermyn et X. Descombes et E. Zhizhina et J. Zerubia. Dans Proc. IEEE SITIS, Publ. IEEE Computer Society, Marrakech, Maroc, décembre 2009. Mots-clés : Extraction d'objets, Processus ponctuels marques, Shape prior, Contour actif, multiple birth-and-death dynamics. @INPROCEEDINGS{Kulikova09a, author = {Kulikova, M. S. and Jermyn, I. H. and Descombes, X. and Zhizhina, E. and Zerubia, J.}, title = {A marked point process model with strong prior shape information for extraction of multiple, arbitrarily-shaped objects}, year = {2009}, month = {décembre}, booktitle = {Proc. IEEE SITIS}, publisher = {IEEE Computer Society}, address = {Marrakech, Maroc}, pdf = {http://hal.inria.fr/docs/00/43/63/20/PDF/PID1054029.pdf}, keyword = {Extraction d'objets, Processus ponctuels marques, Shape prior, Contour actif, multiple birth-and-death dynamics} } Abstract : We define a method for incorporating strong prior shape information into a recently extended Markov point process model for the extraction of arbitrarily-shaped objects from images. To estimate the optimal configuration of objects, the process is sampled using a Markov chain based on a stochastic birth-and-death process defined in a space of multiple objects. The single objects considered are defined by both the image data and the prior information in a way that controls the computational complexity of the estimation problem. The method is tested via experiments on a very high resolution aerial image of a scene composed of tree crowns.

3 - Object extraction from high resolution SAR images using a birth and death dynamics. F. Arslan et X. Descombes et J. Zerubia. Dans Proc. IEEE International Conference on Image Processing (ICIP), Cairo, Egypt, novembre 2009. Mots-clés : High resolution SAR images, Extraction d'objets, Processus ponctuels marques, birth and death process. @INPROCEEDINGS{Fatih09, author = {Arslan, F. and Descombes, X. and Zerubia, J.}, title = {Object extraction from high resolution SAR images using a birth and death dynamics}, year = {2009}, month = {novembre}, booktitle = {Proc. IEEE International Conference on Image Processing (ICIP)}, address = {Cairo, Egypt}, url = {http://dx.doi.org/10.1109/ICIP.2009.5413907}, keyword = {High resolution SAR images, Extraction d'objets, Processus ponctuels marques, birth and death process} } Abstract : We present a new approach to extract predefined objects, such as trees and oil tanks for instance, from high resolution SAR images. We consider a stochastic approach based on an object process also called marked point process. The objects represent trees or oil tanks which are modeled by disks in the image. We first define a Gibbs density that takes into account both prior information and the data. The energy we define is composed of two terms, one is a prior, penalizing overlaps between objects, and the other is a data term, which measures the suitability of an object in the SAR image. The problem is then reduced to an energy minimization problem. We sample the process to extract the configuration of objects minimizing the energy by a fast birth-and-death dynamics, leading to the total number of objects (trees or oil tanks in our case). This approach is much faster than manual counts and does not need any preprocessing or supervision of a user.

4 - 2D and 3D Vegetation Resource Parameters Assessment using Marked Point Processes. G. Perrin et X. Descombes et J. Zerubia. Dans Proc. International Conference on Pattern Recognition (ICPR), Hong-Kong, août 2006. Mots-clés : Energie d'attache aux données, Extraction d'objets, Extraction de Houppiers, Geometrie stochastique, Processus ponctuels marques. @INPROCEEDINGS{perrin_06_c, author = {Perrin, G. and Descombes, X. and Zerubia, J.}, title = {2D and 3D Vegetation Resource Parameters Assessment using Marked Point Processes}, year = {2006}, month = {août}, booktitle = {Proc. International Conference on Pattern Recognition (ICPR)}, address = {Hong-Kong}, pdf = {ftp://ftp-sop.inria.fr/ariana/Articles/2006_perrin_06_c.pdf}, keyword = {Energie d'attache aux données, Extraction d'objets, Extraction de Houppiers, Geometrie stochastique, Processus ponctuels marques} } Abstract : High resolution aerial and satellite images of forests have a key role to play in natural resource management. As they enable to study forests at the scale of trees, it is now possible to get a more accurate evaluation of the forest resources, from which can be deduced information of biodiversity and ecological sustainability. In that prospect, automatic algorithms are needed to give a further exploitation of the data and to assist human operators. In this paper, we present a stochastic geometry approach to extract 2D and 3D parameters of the trees, by modelling the stands as some realizations of a marked point process of ellipses or ellipsoids, whose points are the positions of the trees and marks their geometric features. This approach gives also the number of stems, their position, and their size. It is an energy minimization problem, where the energy embeds a regularization term (prior density), which introduces some interactions between the objects, and a data term, which links the objects to the features to be extracted. Results are shown on aerial images provided by the French National Forest Inventory (IFN).

5 - A comparative study of three methods for identifying individual tree crowns in aerial images covering different types of forests. M. Eriksson et G. Perrin et X. Descombes et J. Zerubia. Dans Proc. International Society for Photogrammetry and Remote Sensing (ISPRS), Marne La Vallee, France, juillet 2006. Mots-clés : Croissance de Region, Processus ponctuels marques, Champs de Markov, Extraction d'objets, Extraction de Houppiers. @INPROCEEDINGS{eriksson06a, author = {Eriksson, M. and Perrin, G. and Descombes, X. and Zerubia, J.}, title = {A comparative study of three methods for identifying individual tree crowns in aerial images covering different types of forests}, year = {2006}, month = {juillet}, booktitle = {Proc. International Society for Photogrammetry and Remote Sensing (ISPRS)}, address = {Marne La Vallee, France}, pdf = {ftp://ftp-sop.inria.fr/ariana/Articles/2006_eriksson06a.pdf}, keyword = {Croissance de Region, Processus ponctuels marques, Champs de Markov, Extraction d'objets, Extraction de Houppiers} } Abstract : Most of today's silviculture methods has the goal to optimise the outcome of the forest in stem volume when it is cut.‭ ‬It might also be relevant to save parts of the forest,‭ ‬for instance,‭ ‬to protect a habitat.‭ ‬In order to get a good survey of the forest,‭ ‬remote sensed images are often used.‭ ‬These images are most often manually interpreted in combination with field measurements in order to estimate the forest parameters that are of importance in the decision how to optimally maintain the forest.‭ ‬Among these parameters the most common are stem number,‭ ‬stem volume,‭ ‬and tree species.‭ ‬Interpretation of images are often labour and time consuming.‭ ‬Thus,‭ ‬automatically developed methods for interpretation can lower the work load and speed up the interpretation time.‭ The interpretation is often done using images captured from a far distance from the ground in order to capture as large area as possible.‭ ‬However,‭ ‬this lower the accuracy of the estimates since it must be done stand wise.‭ ‬Knowledge of where each individual trees in the forest is located together with its size will increase accuracy.‭ ‬It makes it also possible to plan the cutting in detail.‭ ‬With this knowledge in mind,‭ ‬research about finding automatically methods for finding individual tree crowns in aerial images has been a subject for researchers the last decades.‭ Today's methods are not capable to alone handle all kind of forests.‭ ‬Therefore,‭ ‬comparative studies of different segmentation methods with different types of forests are of importance in order to clarify how much a method is reliable at a certain type of forest.‭ ‬This knowledge can,‭ ‬for instance,‭ ‬be used to build up an expert system which are supposed to be able to find individual tree crowns in any kind of forests.‭ ‬The comparison is done using images covering different types of forests.‭ ‬The types of forests that are included in the study ranges from isolated tree crown where the ground is clearly visible between the crowns to dense forest which is naturally regenerated via planted forest. In this study we compare three existing segmentation methods for extracting individual tree crowns from aerial images.‭ ‬The first two methods are probabilistic methods which minimises some energy function while the third is a region growing algorithm.‭ ‬The first probabilistic method is based on a Markov Random Field modelling.‭ ‬We define a prior Markov model to segment the image into three classes‭ (‬background,‭ ‬vegetation and tree centres‭)‬.‭ ‬The prior model embed a circular shape model of the tree crown with a‭ ‬random radius.‭ ‬The data term allows to well position the tree centres onto the image and to describe the tree shape as fluctuations around the circular template.‭ ‬Besides,‭ ‬some long range interactions models the relations between the trees locations,‭ ‬such as some periodicity in case of plantations.‭ The second probabilistic method consists in modeling the trees in the forestry images as random configurations of ellipses or ellipsoids,‭ ‬whose points are the positions of the stems and marks their geometric features.‭ ‬The density of this process embeds a regularization term‭ (‬prior density‭)‬,‭ ‬which introduces some interactions between the objects,‭ ‬and a data term,‭ ‬which links the objects to the features to be extracted.‭ ‬We estimate the best configuration of an unknown number of objects,‭ ‬from which‭ ‬2D and‭ ‬3D vegetation resource parameters can be extracted.‭ ‬To sample this marked point process,‭ ‬we use Monte Carlo dynamics,‭ ‬while the optimization is performed via a Simulated Annealing algorithm,‭ ‬which results in a fully automatic approach.‭ ‬This approach works well on plantations,‭ ‬where there are high spatial relations between the trees,‭ ‬and on isolated trees where‭ ‬3D parameters can be extracted,‭ ‬but some difficulties remain in dense areas.‭ The third method,‭ ‬the region growing algorithm,‭ ‬relies as all region growing methods on good seed points,‭ ‬i.e.‭ ‬in this case approximate locations of the tree crowns.‭ ‬From the seed points the segments are grown according to a grey level value of the neighbouring pixels.‭ ‬The larger the value is the sooner it is connected to the neighbouring segment.‭ ‬The segments stops to grow when all pixels belongs to a segment.‭ ‬This method,‭ ‬contrary the others,‭ ‬will have as a result,‭ ‬segments that have captured the actual shape of the tree crown if the forest is not too sparse.‭ ‬If the forest is too sparse such that the ground is visible,‭ ‬there are problems of finding the seed points.‭ ‬In the cases when the forest is sparse,‭ ‬there are difficulties to separate the tree crowns from the ground.‭ ‬Even if the seed points would be located only at the tree crowns the result will contain a lot of errors since all pixels most belong to a segment,‭ ‬i.e.‭ ‬even the ground pixels must be connected to a segment in this case.‭

6 - Forest Resource Assessment using Stochastic Geometry. G. Perrin et X. Descombes et J. Zerubia et J.G. Boureau. Dans Proc. International Precision Forestry Symposium, mars 2006. Mots-clés : Extraction de Houppiers, Extraction d'objets, Geometrie stochastique, RJMCMC, Energie d'attache aux données. @INPROCEEDINGS{perrin_06_b, author = {Perrin, G. and Descombes, X. and Zerubia, J. and Boureau, J.G.}, title = {Forest Resource Assessment using Stochastic Geometry}, year = {2006}, month = {mars}, booktitle = {Proc. International Precision Forestry Symposium}, pdf = {ftp://ftp-sop.inria.fr/ariana/Articles/perrin_ipfs06.pdf}, keyword = {Extraction de Houppiers, Extraction d'objets, Geometrie stochastique, RJMCMC, Energie d'attache aux données} } Abstract : Aerial and satellite imagery has a key role to play in natural resource management, especially in forestry application. The submetric resolution of the data enables to study forests at the scale of trees, and to get a more accurate assessment of the resources such as the number of stems or the forest cover. To develop automatic tools in order to help the inventories in their work and to bring more knowledge about the stands is also nowadays of important economical and environmental concerns. In this paper, we aim at extracting tree crowns from high resolution aerial Color Infrared images (CIR) of forests using marked point processes. Our approach consists in modelling the trees in the forestry images as random configurations of ellipses, whose points are the positions of the stems and marks their geometric features. The density of this process embeds a regularization term (prior density), which introduces some interactions between the objects, and a data term, which links the objects to the features to be extracted. Our goal is to find the best configuration of an unknown number of objects, i.e. the configuration that maximizes this density. To sample this marked point process, we use Monte Carlo dynamics while the optimization is performed via a Simulated Annealing algorithm, which results in a fully automatic approach. We present different models for the data term in order to cope with different kinds of stands : plantations, isolated trees and mixed stands. Results are shown on aerial CIR images provided by the French Forest Inventory (IFN)

haut de la page

navigation : 1 - 2 - 3 affichage : 10 par page | 20 par page | 50 par page

Ces pages sont générées par