Papers by Gerard Giraudon
HAL (Le Centre pour la Communication Scientifique Directe), Jan 19, 2022
Notre étude se situe dans le domaine du traitement numérique d'image et plus particulièrement... more Notre étude se situe dans le domaine du traitement numérique d'image et plus particulièrement dans celui de l'extraction de contours dans les images. Une des méthodes de détection de contours les plus utilisées est le rehaussement d'image suivi d'un seuillage. C'est une méthode locale qui se fait par la mesure des discontinuités de niveaux de gris sur un voisinage centré, suivi d'une comparaison par rapport à un seuil. Les problèmes de mise en oeuvre d'une telle méthode sont d'une part la synthèse du filtre, et d'autre part le choix de la valeur du seuil. Nous présentons dans cet article, deux nouvelles méthodes complémentaires, une de synthèse générale d'opérateurs différentiels linéaires et non récursifs adaptés aux types de contours, et l'autre permettant le calcul du seuil optimal en fonction du taux de fausses alarmes désiré. Après avoir explicité la méthode de manière théorique, celle-ci est illustrée et comparée aux méthodes existantes, sur des images de scènes naturelles. L'avantage et l'originalité de cette méthode par rapport à celles existantes est de permettre la sélection a priori du type de signal à détecter ou à rejeter comme faisant partie d'un contour. Elle opère donc de manière sélective sur les contours contenus dans une image, la détection se faisant à seuil constant quelque soit le rapport S/B sur l'image, à taux de fausses alarmes fixé
HAL (Le Centre pour la Communication Scientifique Directe), Dec 10, 2020
Préface Résumé Recommandation n° 7 : Créer un observatoire des EdTechs 7. Annexes Annexe 1 : Comp... more Préface Résumé Recommandation n° 7 : Créer un observatoire des EdTechs 7. Annexes Annexe 1 : Compléments sur l'intelligence artificielle Annexe 2 : Quelques exemples de travaux de recherche des équipes-projets Inria Annexe 3 : Exemples d'initiatives pour la formation au numérique
Smoothing (regularization), interpolation and surface reconstruction are well known subjects in c... more Smoothing (regularization), interpolation and surface reconstruction are well known subjects in computer vision. The major difficulty is to choose a model well suited for one of these goals and driven by a minimum number of parameters. Another problem also arises when we want to do one of these opemtiow adaptatively, i.c. local features are processed in keeping m'th the application domain (e.g. cartography). Our goal is to present a discrete operator driven by only one parameter, allowing both global and local prvcessing of a surface and, well suited to smoothing, interpolation and surface reconstruction.
Fractals, Sep 1, 1994
As for almost all physical signals, useful information for image understanding is contained in th... more As for almost all physical signals, useful information for image understanding is contained in the position and nature of singularities. Consequently, a set of early vision methods has been proposed for locating discontinuities of given derivatives with the aim of solving a large class of features extraction problems. Certain early vision tasks also require a more accurate knowledge of pointwise regularity, mainly for evaluating image roughness. We propose to take these requirements into account, by extending the usual singularity detection methods with the help of fractional calculus.
Proceedings of SPIE, Aug 17, 1994
Isprs Journal of Photogrammetry and Remote Sensing, Feb 1, 1997
Our paper presents an automatic generation of high resolution urban digital elevation models (DEM... more Our paper presents an automatic generation of high resolution urban digital elevation models (DEMs) based on a highly redundant correlation process. We will discuss the difficulties of such a task by commenting on the state of the art, and we propose an approach in three main steps. In the first step, the image acquisition specification as image sequences leads to pairs with various base/height ratios in order to obtain good precision and few errors due to hidden parts. In the second step we use various stereovision methods and we merge the results, thus attributing to each pixel the most probable and precise elevation. In the third step we automatically extract terrain-DEM and building-DEM from computed DEM in order to specifically post-process each class. Finally, we combine these two DEMs to generate a final DEM which presents the best continuity for ground surface, and which respects sharp building discontinuities. The results obtained with an operational example (including image size, difficulty of the scene) demonstrate the feasibility of generating metric resolution urban data bases from automated digital stereo methods.
HAL (Le Centre pour la Communication Scientifique Directe), Nov 7, 2012
Bulletin 1024, Nov 1, 2021
Lecture Notes in Computer Science, 1996
ABSTRACT The rate of information transmission of a transmitter-receiver couple was defined by C.E... more ABSTRACT The rate of information transmission of a transmitter-receiver couple was defined by C.E. Shannon in his Information Theory. We use this concept in Computer Vision to define models of image redundancy. Houzelle et Giraudon applied information theory to a simple model considering an image as a set of isolated pixels. We introduce a Markov Random Field model to take into account the spatial neighbourhood of a pixel. We show that we have to determine some parameters of the MRF in order to obtain sufficient statistics from common satellite images, and we propose a measure based on a generalized Ising model. Another model which also takes into account a pixel's spatial context is then proposed. It considers the correspondence between grey level vectors of cliques. We introduce a distance in the grey level space to solve the problem of insufficient statistics. Finally, results for the proposed definitions are presented for some synthetic and a large variety of SPOT XS1, XS2 and XS3 image triples and are compared to the classical correlation coefficient measure.
The architecture of an image analysis system called MESSIE (Multi Expert System for Scene Interpr... more The architecture of an image analysis system called MESSIE (Multi Expert System for Scene Interpretation and Evaluation) is presented. This system reasons from geometric models which are represented by four concepts (geometry, radiometry, context, and function). The aim is to find the class instances from the generic models which are present in the scene. The necessity of having a hierarchic
Springer eBooks, 1994
This paper is devoted to an analytical study of extrema curvature evolution through scale-space. ... more This paper is devoted to an analytical study of extrema curvature evolution through scale-space. Our analytical study allows to get results which show that, from a qualitative point of view, corner evolution in scale-space has the same behavior for planar curves or surfaces. In particular, this analysis, performed with different corner-shape models, shows that, for a two-corner shape, two curvature maxima exist and merge at a certain scale er0, depending on the shape. For a two-corner grey-level surface, the evolution of the determinant of hessian (DET) shows a merging point for a certain a0 independently of contrast, and the evolution of Ganssian Curvature presents the same characteristic but this point evolves with contrast.
Robotics and Autonomous Systems, Oct 1, 1997
This paper presents MESSIE, a multi-specialist architecture for scene interpretation in a robotic... more This paper presents MESSIE, a multi-specialist architecture for scene interpretation in a robotic application. MESSIE is a centralized hierarchical blackboard architecture. The generic model of objects and the explicit description of sensors and materials allow the use of an application independent interpretation strategy. Two remote sensing applications on 2D scene interpretation and a third one, presented in this paper, on 3D scene interpretation allow the validation of the proposed architecture as well as the main features of MESSIE. After a brief overview of the state of art in 3D object modeling and scene interpretation, we discuss the scene interpretation problem from the knowledge representation view point. Then the architecture of MESSIE, the object modeling and the processing strategies (object detection and scene interpretation) are described. Further an application of 3D indoor scene interpretation in mobile robot context is given. We also present an interpretation running example using constrained low-level feature extraction mechanism to improve the image segmentation results.
Lecture Notes in Computer Science, 1990
“Scene Analysis”, especially for real data, is a complex problem. There are two main explanations... more “Scene Analysis”, especially for real data, is a complex problem. There are two main explanations which interest us in this paper.
A region segmentation algorithm is presented, using a model for joint probability density. Joint ... more A region segmentation algorithm is presented, using a model for joint probability density. Joint probability density can be defined as an N×N cooccurrence matrix in which each coordinate (i, j) gives the probability for the gray-level transition i, j between two neighbor pixels. The approach consists in modeling the energy distribution within a cooccurrence matrix of a region. Regions are
IEEE Transactions on Geoscience and Remote Sensing, Jul 1, 1999
OPSILA, is a description given of a general-purpose parallel architecture with two different form... more OPSILA, is a description given of a general-purpose parallel architecture with two different forms of parallelism: the well-known SIMD (single-instruction, multiple data shown), a synchronous form of parallelism; and SPMD (single program, multiple data stream), which is an asynchronous mode. It is shown that OPSILA is efficient for a wide variety of image algorithms including low and high level processing.
HAL is a multidisciplinary open access archive for the deposit and dissemination of scientific re... more HAL is a multidisciplinary open access archive for the deposit and dissemination of scientific research documents, whether they are published or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers. L'archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d'enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.
TS. Traitement du signal, 1989
Ce travail s'inscrit par le cadre du développement méthodologique d'une architecture multi spécia... more Ce travail s'inscrit par le cadre du développement méthodologique d'une architecture multi spécialistes, appelée MESSIE, pour l'interprétation d'images. Nous présentons, dans cet article, les spécifications du système pour l'interprétation d'objets, en particulier ceux faits de main d'homme dans l'imagerie aérienne. La principale difficulté de ce problème est l'expression des connaissances nécessaires à l'interprétation : les connaissances liées aux Objets, à la Stratégie et à la Scène. Nous montrons comment une modélisation des objets physiques de la scène, exprimée sous quatre points de vue (géométrie-radiométrie (forme-aspect), relation spatiale entre objets (contexte) et fonction (sémantique)) permet d'aboutir volume 9-n° 5 à la spécification d'une architecture. L'architecture MESSIE('), de type tableau noir, est organisée sur ces quatre types de bases de connaissances, qui schématiquement se décomposent en deux niveaux. Le premier niveau correspond à la connaissance liée à la Scène et à la Stratégie et le deuxième correspond aux spécialistes (un par type d'objet). Chaque niveau n'accède qu'à certains des points de vue. Nous montrons la modularité d'une telle approche et la facilité de mise en place de diverses This work is in the context of the methodological development of a multispecialist architecture called MESSIE. This paper presents the system specifications for the interpretation of man-made structures in the field of aerial imagery. The main difficulty of such a system is the knowledge expression necessary for the interpretation : among them, strategy, scene and objects knowledges. We show how the choosen modeling of physical objects described by four points of view allows to lead to the specifieation of an architecture. These points of view are geometry and radiometry modeling (shape-aspect), spatial relation of the objet with the others (context) and the functionnality (semantic). MESSIE is a blackboard architecture organized with four types of knowledge bases that schematically are grouped in two hierarchical levels. The first deux domaines ci-dessus pour la réalisation de l'application. (1) This work is supported by the Esprit Project 1588. Traitement du Signal ystèmes experts Un système multi spécialistes en vision ABSTRACT 404 stratégies et nous illustrons l'article avec les résultats obtenus par quatre spécialistes d'objets artificiels (routes, bâtiments, ombres et voitures) et par un spécialiste d'objet naturel (cours d'eau). MOTS CLÉS Vision par Ordinateur, Interprétation d'images, images aériennes. level corresponds to the Scene and Strategy level and the second corresponds to the specialists (one for each objet). Each level works only with certain points of view. We show the modularity of a such approach and the facility to use différent strategies. We illustrate this paper with results obtained with four specialists : Road, Building, Shadow and Vehicule for man-made objects and one specialist for natural object (river) .
Uploads
Papers by Gerard Giraudon