Dissertations / Theses on the topic 'Traitement d'images 3D'
Create a spot-on reference in APA, MLA, Chicago, Harvard, and other styles
Consult the top 50 dissertations / theses for your research on the topic 'Traitement d'images 3D.'
Next to every source in the list of references, there is an 'Add to bibliography' button. Press on it, and we will generate automatically the bibliographic reference to the chosen work in the citation style you need: APA, MLA, Harvard, Chicago, Vancouver, etc.
You can also download the full text of the academic publication as pdf and read online its abstract whenever available in the metadata.
Browse dissertations / theses on a wide variety of disciplines and organise your bibliography correctly.
Desbarats, Pascal. "Structuration d'images segmentées 3D discrètes." Bordeaux 1, 2001. http://www.theses.fr/2001BOR12486.
Full textItthirad, Frédéric. "Acquisition et traitement d'images 3D couleur temps réel." Thesis, Saint-Etienne, 2011. http://www.theses.fr/2011STET4011.
Full textThe existing 3D sensors aren’t much used and are only capable of capturing 3D dat. When 2D data are necessary, one has to use another camera and correlate the 2 images. NT2I has decided to develop its own solution in order to control the acquisition chain. My work has been to develop a specific camera with color, calibration, and image processing algorithms. In that purpose, I've worked on the extension of the LIP model (Logarithmic Image Processing) for color images and on the implementation of real time algorithms
Voisin, Sophie. "3D model acquisition, segmentation and reconstruction using primitive fitting." Dijon, 2008. http://www.theses.fr/2008DIJOS056.
Full textThe reverse engineering of a 3D object consists to identify the main parts or primitives, which best reconstruct its 3D point cloud. Because the success of the reconstruction process is greatly influenced by the errors generated along the reverse engineering chain, we focus our research on improving two phases of the process. Firstly, in order to minimize the point cloud acquisition errors associated with the use of a structured light projection scanner, we present a method to select the best illumination source and the best object appearance colors depending on the characteristics of the scanner used. Secondly, in order to obtain a simplified representation of the object while maintaining accuracy and realistic representation, we present novel 3D reconstruction and segmentation methods. The originality of these methods is the use of genetic algorithms to obtain the representation of the model using primitives, in our case using superquadriques or supershapes. The particularities of these methods lie in the flexibility provided by the genetic algorithms in solving optimization problems since they do not depend on the initialization process, and lie on the capabilities of the supershapes representation allowing to reconstruct very complex 3D shapes. Despite computing time relatively expensive, we present good performance results in terms of reconstruction and segmentation of objects and/or scenes
Rigaudière, Dominique. "Reconstruction et modélisation 3D à l'aide de squelettes et de surfaces implicites." Dijon, 2002. http://www.theses.fr/2002DIJOS038.
Full textRipoche, Xavier. "Méthodes de traitement d'images radiologiques 3D : applications thérapeutiques et chirurgicales." Paris 11, 2006. http://www.theses.fr/2006PA112040.
Full textBonnard, Jennifer. "Génération d'images 3D HDR." Thesis, Reims, 2015. http://www.theses.fr/2015REIMS014/document.
Full textHDR imaging and 3D imaging are two areas in which the simultaneous but separate development has been growing in recent years. On the one hand, HDR (High Dynamic Range) imaging allows to extend the dynamic range of traditionnal images called LDR (Low Dynamic Range). On the other hand, 3Dimaging offers immersion in the shown film with the feeling to be part of the acquired scene. Recently, these two areas have been combined to provide 3D HDR images or videos but few viable solutions existand none of them is available to the public. In this thesis, we propose a method to generate 3D HDR images for autostereoscopic displays by adapting a multi-viewpoints camera to several exposures acquisition.To do that, neutral density filters are fixed on the objectives of the camera. Then, pixel matchingis applied to aggregate pixels that represent the same point in the acquired scene. Finally, radiance is calculated for each pixel of the set of images by using a weighted average of LDR values. An additiona lstep is necessary because some pixels have wrong radiance. We proposed a method based on the color of adjacent pixels and two methods based on the correction of the disparity of those pixels. The first method is based on the disparity of pixels of the neighborhood and the second method on the disparity independently calculated on each color channel. This pipeline allows the generation of 3D HDR image son each viewpoint. A tone-mapping algorithm is then applied on each of these images. Their composition with filters corresponding to the autostereoscopic screen used allows the visualization of the generated 3DHDR image
Cinquin, Philippe. "Application des fonctions-spline au traitement d'images numériques." Habilitation à diriger des recherches, Grenoble 1, 1987. http://tel.archives-ouvertes.fr/tel-00325721.
Full textLubniewski, Pawel. "Recalage 3D/2D d'images pour le traitement endovasculaire des dissections aortiques." Thesis, Clermont-Ferrand 1, 2014. http://www.theses.fr/2014CLF1MM24/document.
Full textIn this study, we present our works related to 3D/2D image registrationfor aorti dissition. Its aim is to propose a visualization of medial datawhih an be used by physians during endovas ular proedures.For this purpose, we have proposed a parametrimodel of aorta, alleda Tubular Envelope. It is used to express the global shape and deformationsof the aorta, by a minimal number of parameters. The tubular envelope isused in our image registration algorithms.The registration by ITD (Image Transformation Descriptors) is our ori-ginal method of image alignment : itomputes the rigid 2D transformation between data sets diretly, without any optimization process.We provide thedefinition of this method, as well as the proposition of several descriptors' formulae, in the base of images of aorta. The technique allows us to quickly and a poarse alignment between data. We also propose the extension of theoriginal approach for the registration of 3D and 2D images.The complete chain of 3D/2D image registration techniques, proposedin this document, consists of the ITD stage, followed by an intensity basedhybrid method. The use of our 3D/2D algorithm, based on the image trans-formation descriptors as an initialization phase, reduces the computing timeand improves the efficiency of the presented approach.We have tested our registration methods for the medical images of several patients after endovasular treatment. Results have been approved by our clinical specialists and our approach.We have tested our registration methods for the medical images of several patients after endovascular treatment. Results have been approved by our clinical specialists and our approach may appear in the intervention rooms in the futur
El-Hajjaji, Abdellah. "Traitement numérique en 3D d'un couple d'images stéréo du satellite SPOT." Rouen, 1993. http://www.theses.fr/1993ROUES028.
Full textThe aim of our research was to extract the level h of a precise landscape taken over two differents angles by the satellite SPOT. To do so, we have modelised the movement of the satellite and his optical system to transform the two first images in another one, epipolar which will allow us to reduce the matching time and to find with success the equivalent pixels. For the pairing, we have utilised a technic wich is based on the corrolation and of the dynamic programming. This method was very satisfactory and allow us to match 96 % of the equivalent pixels, with an error of less than 5 meters, but the original problem is still a matter of research for complimentary studing
Peyrin, Françoise. "Méthodes de reconstruction d'images 3D a partir de projections coniques de rayons X." Lyon, INSA, 1990. http://www.theses.fr/1990ISAL0028.
Full text[This work is concerned with truly X-ray 3D tomography. The method consists in the acquisition of radiographs of an object for different positions of a X-ray cone beam source. Then the Image is obtained by solving a 3D reconstruction from cone-beam projection problem. The method is first situated with respect to the other imaging modalities in tomography. After a bibliographic study of reconstruction technicals, the generalisation of the back-projection theorem for cone-beam projection is proved. It leads to an inversion formula, which is exact when the source describes the surface of a sphere. The subsequent algorithm iS implemented on a super calculator and tested on simulations. When the source trajectory satisfies the theorem hypothesis, the reconstructed images are satisfying. Other while they can be considered as a first approximation of the object. In order to be less sensitive to the acquisition conditions, an algebraic approach to this problem is considered. The use of three bloc Iterative reconstruction algorithms including or not some simple constraints is proposed. At last, the different methods are applied to the reconstruction of a bony phantom from physical projections, and some conclusions concerning the realisation of a truly 3D X-ray scanner are presented. ]
Le, Troter Arnaud Alexandre Alain. "Mise en correspondance d'images et de modèles : application à la reconstruction 3D de scènes sportives." Aix-Marseille 2, 2006. http://theses.univ-amu.fr.lama.univ-amu.fr/2006AIX22045.pdf.
Full textThis PhD is part of the SimulFoot project, whose main goal is the analysis of soccer scenes. Our research deals with image sequence analysis, geometrical modeling, and visualization, as well as simulation and cognitive studies. Our work focuses on image processing methods for automatically reconstructing a 3D model from a single image in the scene. A first step provides the detection of a region of interest (the field) by colorimetric analysis. Our method is illustrated with the study of pixel distribution in the HLS color space. We then propose a novel method for color space categorization at the perceptual level in CIELab space. A second step proposes an original approach for extracting (using a Hough transform) the landmarks of a region of interest (straight lines and ellipses) and matching them with the corresponding features in the 3D model
Burg, Samuel. "Segmentation 3D d'images scintigraphiques et simulations très réalistes GATE." Poitiers, 2011. http://nuxeo.edel.univ-poitiers.fr/nuxeo/site/esupversions/b5bc3be2-14be-4526-9e11-61619c11caaf.
Full textThe objective of this thesis was to propose a new 3D segmentation method for scintigraphic imaging. The first part of the work was to simulate 3D volumes with known ground truth in order to validate a segmentation method over other. MonteCarlo simulations were performed using the GATE software (Geant4 Application for Emission Tomography). For this, we characterized and modeled the gamma camera "Imager" Biospace™ by comparing each measurement from a simulated acquisition to his real equivalent. The "low level" segmentation tool that we have developed is based on a modeling of the levels of the image by probabilistic mixtures. Parameters estimation is done by an SEM algorithm (Stochastic Expectation Maximization). The 3D volume segmentation is achieved by an ICM algorithm (Iterative Conditional Mode). We compared the segmentation based on Gaussian and Poisson mixtures to segmentation by thresholding on the simulated volumes. This showed the relevance of the segmentations obtained using probabilistic mixtures, especially those obtained with Poisson mixtures. Those one has been used to segment real 18FDG PET images of the brain and to compute descriptive statistics of the different tissues. In order to obtain a "high level" segmentation method and find anatomical structures (necrotic part or active part of a tumor, for example), we proposed a process based on the point processes formalism. A feasibility study has yielded very encouraging results
MIGEON, BRUNO. "Reconstruction 2d et 3d d'os longs des membres par traitement d'images echographiques." Orléans, 1993. http://www.theses.fr/1993ORLE2025.
Full textDebon, Renaud. "Analyse d'images échographiques de loesophage, reconstruction 3D et interprétation." Rennes 1, 2005. http://www.theses.fr/2005REN1S196.
Full textThis work concerns the approach evaluations, which are oriented « knowledge based system » for medical images interpretation applied to esophagus echoendoscopy and the development of aid system for tumor staging. It's shown how the intelligent approaches (expert system and information fusion) can allow rationalizing the using of a priori knowledge. The pertinent extraction of anatomic structures, in our case, esophagus structure, becomes a natural application in the knowledge engineering. This extraction is based on the image segmentation. The required robustness for these algorithms impose the advanced architectures development allowing the compensation of low numerical content of these images. Three concrete examples are detailed : 2D extraction of the esophagus' interface, 3D extraction of the esophagus' interfaces and spatial following with 3D reconstruction of the aorta. Knowledge is represented by static or dynamical model (fuzzy, geometric or evidential models). The approach using the complementarities of probabilities and fuzzy logic to obtain presentation « exact » of knowledge a priori. Fuzzy models and statistic reality are synchronized by a knowledge base. It's shown that all components can be integrated in a coherent architecture hierarchically organized
Chebaro, Bilal. "Etude du mouvement 3D d'objets par l'analyse d'une séquence stéréoscopique d'images." Toulouse 3, 1993. http://www.theses.fr/1993TOU30195.
Full textBaere, Campos Neves José Alberto. "Contribution à la construction automatique de représentation 3D d'objets solides." Compiègne, 1989. http://www.theses.fr/1989COMPD171.
Full textDouadi, Lounis. "Contribution à l'étude du recalage de données 3D /couleur." Montpellier 2, 2006. http://www.theses.fr/2006MON20111.
Full textGuerchouche, Rachid. "Analyse multi-vues d'objets 3D pour interactions collaboratives." Evry, Institut national des télécommunications, 2008. http://www.theses.fr/2008TELE0022.
Full textThe objective of this thesis is the 3D reconstruction of real objects for collaborative interactions. Within this framework, the goal is to reconstruct an object from a small number of calibrated views (8 to 12 images), and then insert the obtained numerical models into shared collaborative environments for further visualization and manipulation. The thesis is organized as follows. First, in the general introduction, we introduce the context of the research presented in this work. We present the acquisition system developed within our laboratory, and show the relationship between the calibration of this system, the image acquisition process, and the collaborative interactions with the reconstructed objects. Then, in a first chapter, we present some aspects of 3D geometry applied to computer vision. We present the projective geometry theory, the classical linear and non-linear camera models, and some stereoscopic vision results useful for the rest of the document. The second chapter is dedicated to the issue of camera calibration. After a state-of-the-art of camera calibration methods, we propose a robust camera calibration method based on the robust estimation of the perspective projection matrix. The calibration pattern used is a cube with faces of different colours. The proposed calibration algorithm uses one image per camera to perform the calibration. However, in order to increase the accuracy of the camera parameters estimation, multiple images can also be used. Our method yields a robust estimation of the camera parameters while minimizing the amount of user interaction requested. In order to validate the method, we introduce a set of new objective criteria for evaluation and comparison of camera calibration methods. The proposed criteria are based on rectification and 3D reconstruction of an unknown coplanar point set, a virtual pattern, and the re-estimation of the known parameters of stereoscopic systems. Our calibration method is finally validated according to the proposed criteria. The third chapter3D tackles the issue of 3D reconstruction of real objects. After a comprehensive state-of-the-art of 3D reconstruction methods, we present our proposed multiresolutiuon 3D reconstruction algorithm, which is adapted to collaborative interaction tasks. Our contributions specifically concern new algorithms for voxel visibility and photo-consistency estimation. The proposed 3D reconstruction method is then tested and validated upon a set of images of real objects from existing benchmark databases. The fourth chapter handles the collaborative interactions with 3D objects. First, the calibration of our acquisition system composed of eight cameras is presented. Experimental results concerning 3D reconstruction of objects available in our laboratory are then presented. Finally, collaborative interactions with the reconstructed objects are illustrated within the framework of three existing interfaces in the France Telecom R\&D laboratories: MOWGLI, DigiTable and Spin3D. A concluding section summarizes the contributions of this thesis and opens perspectives of future work
Lamallem, Abdellah. "Évaluation de performance d'un système de fusion d'informations : application au traitement d'images 3D." Phd thesis, Université de Grenoble, 2012. http://tel.archives-ouvertes.fr/tel-00768212.
Full textDeschênes, François. "Estimation des jonctions, du mouvement apparent et du relief en vue d'une reconstruction 3D de la scène." Sherbrooke : Université de Sherbrooke, 2000.
Find full textGalisot, Gaëtan. "Segmentation incrémentale et interactive d'images médicales 3D." Thesis, Tours, 2018. http://www.theses.fr/2018TOUR4035.
Full textThis research work describes a new interactive and incremental method for the segmentation of 3D medical images. The a priori information associated to the anatomical structure to analyze is leamed in a local way. Several local atlases, each one describing only one anatomical structure are constwcted from a training dataset. Spatial relationships are also leamed between those regions aiming to position the local atlases inside the whole image. During the segmentation process, the graph is used in an incremental way allowing fast and partial segmentation. fle user can also interact during the local atlas posiboning in order toimprove the segmentation quality. A voxel classification by a hidden Markov random field is employed toprovide the local segmentations. We also propose s post-processing step in order to correct the systematiceuors that a segmentation can achieve
Buchholz, Bert. "Abstraction et traitement de masses de données 3D animées." Phd thesis, Télécom ParisTech, 2012. http://pastel.archives-ouvertes.fr/pastel-00958339.
Full textDupas, Alexandre. "Opérations et Algorithmes pour la Segmentation Topologique d'Images 3D." Phd thesis, Université de Poitiers, 2009. http://tel.archives-ouvertes.fr/tel-00466706.
Full textBuchholz, Bert. "Abstraction et traitement de masses de données 3D animées." Electronic Thesis or Diss., Paris, ENST, 2012. http://www.theses.fr/2012ENST0080.
Full textIn this thesis, we explore intermediary structures and their relationship to the employed algorithms in the context of photorealistic (PR) and non-photorealistic (NPR) rendering. We present new structures for rendering as well as new uses for existing structures. We present three original contributions in the NPR and PR domain: First, we present binary shading, a method to generate stylized black and white images, inspired by comic artists, using appearance and geometry in a graph-based energy formulation. The user can control the algorithm to generate images of different styles and representations. The second work allows the temporally coherent parameterization of line animations for texturing purposes. We introduce a spatio-temporal structure over the input data and an energy formulation for a globally optimal parameterization. Similar to the work on binary shading, the energy formulation provides a an important and simple control over the output. Finally, we present an extension to Point-based Global Illumination, a method used extensively in movie production during the last years. Our work allows compressing the data generated by the original algorithm using quantification. It is memory-efficient and has only a neglegible time overhead while enabling the rendering of larger scenes. The user can easily control the strength and quality of the compression. We also propose a number of possible extensions and improvements to the methods presented in the thesis
Deschamps, Thomas. "Extractions de courbes et surfaces par méthodes de chemins minimaux et ensembles de niveaux : applications en imagerie médicale 3D." Paris 9, 2001. https://portail.bu.dauphine.fr/fileviewer/index.php?doc=2001PA090038.
Full textIn this thesis, we focus on the use of minimal path techniques and Level-Sets active contours, for curve and shape extraction in 3D medical images. In the first part of thesis, we worked upon the reduction of the computing cost for path extraction. We proposed several path extraction algorithms for 2D as well as for 3D images. And we applied those techniques to real medical imaging problems, in particular automatic path extraction for virtual endoscopy and interactive and real-time path extraction with on-the-fly training. In the second part, we focused on surface extraction. We developed a fast algorithm for pre-segmentation, on the basis of the minimal path formalism of the first part. We designed a collaborative method between this algorithm and a Level-Sets formulation of the problem, which advantage is to be able to handle any topological change of the surfaces segmented. This method was tested on different segmentation problems, such as brain aneurysms and colon polyps, where target is accuracy of the segmentation, and enhanced visualization of the pathologies. In the last part of the thesis, we mixed results from previous part to design a specific method for tubular shape description and segmentation, where description is the extraction of the underlying skeleton of our objects. The skeletons are trajectories inside our objects, which are used as well for virtual inspection of pathologies, as for accurate definition of cross-sections of our tubular objects. In the last chapter we show applications of our algorithms to the extraction of branching structures. We study the vascular tree extraction in contrast enhanced medical images, and we apply the same principle to the more complex problem of the bronchial tree extraction in multi-slice CT scanners of the lungs
Merad, Djamel. "Reconnaissance 2D/2D et 2D/3D d'objets à partir de leurs squelettes." Evry-Val d'Essonne, 2004. http://www.theses.fr/2004EVRY0029.
Full textGuerfi, Ababsa Souhila Lelandais Bonadè Sylvie. "Authentification d'individus par reconnaissance de caractéristiques biométriques liées aux visages 2D/3D." S. l. : Evry-Val d'Essonne, 2008. http://www.biblio.univ-evry.fr/theses/2008/2008EVRY0032.pdf.
Full textSouami, Feryel. "Traitement de la réflectance des objets dans un système d'acquisition 3D couleur /." Paris : Ecole nationale supérieure des Télécommunications, 1994. http://catalogue.bnf.fr/ark:/12148/cb35723475d.
Full textCoudert, Sébastien. "Mesures 3D par caméras CCD de champs de vitesse dans des écoulements turbulents." Saint-Etienne, 2002. http://www.theses.fr/2002STET4002.
Full textThis work consists with the developments and the improvements of experimental techniques that measure 3D velocity within flow of fluid. These techniques are stereoscopic particle image velocimetry (SPIV) systems in angular configuration. . . [etc. ]
Espiau, François-Xavier. "Métrologie 3D par vision active sur des objets naturels sous-marins." Nice, 2002. http://www.theses.fr/2002NICE5713.
Full textKhouas, Leila. "Visualisation de champs de vecteurs 2D et 3D par modélisation autorégressive bidimensionnelle d'une texture de type fourrure." Lyon, INSA, 1998. http://www.theses.fr/1998ISAL0081.
Full textIn many areas such as fluid dynamics or medical imaging, the analysis of studied phenomena produces complex data that consists in large vector fields. The vectors represent some characteristic of each point on the field such as: a fluid vorticity, a wind velocity or a motion speed. The visualization of vector fields is not straightforward because they have no natural representation. In this work, we propose to build a new representation of vector fields based on furlike texture. We assume that such a texture provides a natural and intuitive representation of a dense vector field. Our approach consists in the development of a texture model that allows 2D synthesis of furlike texture having a 3D aspect. The model is based on the modeling of the texture autocorrelation function (ACF) and a two dimensional Auto Regressive synthesis (2D AR). This provides a simple and efficient 2D generator of furlike texture with a local control of the main attributes of the texture (orientation and length of filaments). We have experimented the use of this texture model to represent vector fields. We use orientation, length and color attributes of our furlike texture to visualize local orientation and magnitude of a 2D vector field. Results are presented using simulated data and cardiac imaging data. We also show how to visualize 3D vector fields defined over 3D surfaces by a simple and appropriate texture mapping procedure. We have applied this technique for special vector fields that simulate fur appearance on 3D objects. This produces images with quite realistic aspect
Bosnjak, Seminario Antonio. "Segmentation et modélisation dynamiques : application à la reconstruction 3D d'images échocardiographiques." Rennes 1, 2003. http://www.theses.fr/2003REN10018.
Full textRahel, Rahel. "Mesure de champs de déplacements 3D à l'aide d'un scanner laser couplé à une caméra." Dijon, 2009. http://www.theses.fr/2009DIJOS033.
Full textAl, Alwani Adnan Salih. "Event and action recognition from thermal and 3D depth Sensing." Caen, 2016. http://www.theses.fr/2016CAEN2003.
Full textModern computer vision algorithms try to understand the human activity using 3D visible sensors. However, there are inherent problems using 2D visible sensors as a data source. First, visible light images are sensitive to illumination changes and background clutter. Second, the 3D structural information of the scene is degraded when mapping the 3D scene to 2D images. Recently, the easy access to the RGBD data at real-time frame rate is leading to a revolution in perception and inspired many new research. Time of Flight (ToF) and multi-view sensors have been used to model the 3D structure of the scene. Otherwise, infrared thermography (IRT), also known as thermal imaging, is an ideal technology to investigate thermal anomalie under different circumstances because it provides complete thermal images of an object with no physical attachments (nonintrusive). IRT is now being introduced to a wide range of different applications, such as medical diagnostic and surveillance. However, finding meaningful features from a time series data from thermal video is still a challenging problem, especially for event detection. This problem is particularly hard due to enormous variations in visual and motion appearance of object, moving background, occlusions and thermal noise. In this thesis, we propose a framework for the detection of visual events in thermal video and 3d human actions in RGBD data. Despite differences in the applications, the associated fundamental problems share numerous properties, for instance the necessity of handling vision-based approach for the automatic recognition of events. The first part of the thesis deals with the recognition of events in thermal video. In this context, the use of time series is challenging due to the graphical nature which exposes hidden patterns and structural changes in data. In this study, we investigated the use of visual texture patterns for time series classification. Our principal aim was to develop a general framework for time series data mining based on event analysis with an application to the medical domain. In particular, we are interested to pain/no-pain detection using parametric statistics and shape descriptors in order to analyze and to classify time 2D distribution data sets. We first extracted automatically thermal-visual facial features from each face considered as the region of interest (ROI) of the image. We proposed two feature descriptors for the signal pattern of interest (POI) which efficiently exploits the dependence between time and frequency in one-dimension (1D) signal. The original signal is extracted directly from local patch in ROI. The first method is based on non-redundant temporal local binary pattern (NRTLBP). The second approach propose a topological persistence descriptor (TP) for extracting and filtering local extrema of 1D signal. Local minima and local maxima are extracted, paired, and sorted according to their persistence. The final representation of an event is a completely new feature vector of all paired critical values. These features provide many benefits for many applications to get a fast estimation of the event in dynamic time series data. Both methods are validated using an Extreme Learning Machine (ELM) and Support vector Machine (SVM) classifiers. Experimental results on a real thermal-based data set ”Pain in Preterm Infants” (PPI), which is captured in a real condition monitoring environment, show that the proposed methods successfully capture temporal changes in events and achieve higher recognition rates. PPI dataset was developed in the context of Infant pain project, a french project supported by the French National Research Agency Projects for science (ANR). _ In the second part of the thesis, we investigate the problem of recognizing human activities in different application scenarios: controlled video environment( e. G. Indoor surveillance) and specially depth or skeletal data (e. G. Captured by Kinect). We focus on developing spatio-temporal features, and applying these features to identify human activities from a sequence of RGB-D images, i. E. ,color images with depth information. First, we proposed a view-invariant approach which use joint angles and relative joint positions as features. These features are quantized into posture visual words and their temporal transitions are encoded as observation symbols in a Hidden Markov Model (HMM). To eliminate rotation dependence in skeletal descriptors, we proposed an approach that combines the covariance descriptor and the spherical harmonics (SHs). The harmonic representation of 3d shape descriptors is adapted to skeleton joint-based human action recognition. To improve the accuracy and the convergence speed of the SHs solutions, we proposed an extension of the model, using quadratic spherical harmonics (QSH) representation, to encode pose information in the spatiotemporal space. These SHs representations are compact and discriminating. For the recognition task, we used ELM classifier. Our experimental results on a number of popular 3d action datasets show significant achievements in terms of accuracy, scalability and efficiency in comparison to alternate methods, of the state-of-the-art
Oudinet, Ghislain. "Analyse d'images et modélisation 2D/3D de la microstructure du combustible MOX." Saint-Etienne, 2003. http://www.theses.fr/2003STET4011.
Full textThe microstructure of the MOX fuel, made with UO2 and PuO2, determines his " in pile " behavior. The french companies CEA and COGEMA are highly interested in its description by image analysis, which is the object of the present work. The segmentation algorithms described here use pictures issued from a microprobe and a SEM, to analyse the plutonium and porosity distribution in the fuel pellets. They are innovating, automated and robust enough to be used with a small data set. They have been successfully tested on different fuels, before and after irradation. Three-dimensional informations have been computed with a genetic algorithm. The obtained 3D object size distributions allowed the modeling of many different industrial and research fuels. 3D reconstruction is accurate and stable, and provides a basis for different studies among which the study of the MOX fuel " in pile " behavior
Bonnet, Stéphane. "Approches multi résolution en reconstruction tomographique 3D : Application à l'angiographie cérébrale." Lyon, INSA, 2000. http://www.theses.fr/2000ISAL0077.
Full text3D rotational angiography (RA) is raising increasing interest for diagnostic in the field of endovascular treatment of intracranial Aneurysms. It brings to the physician valuable 3d information of the cerebral vascular 3D reconstruction of arterial vessels is Obtained via a cone-beam reconstruction algorithm applied onto a set of contrast-enhanced x-ray images. This thesis suggests new Tomographic reconstruction methods in order to access directly the Representation of a 3d object at different resolutions from its projections. The underlying idea is to apply image processing techniques in this intermediate space to increase both image quality and Computation time. In a first step, the natural relationship between wavelets and Computerized tomography is investigated for parallel beam geometry. It provides a rigorous mathematical framework where multiresolution Analysis can be considered. We present an original tomography Reconstruction algorithm satisfying either separable or quincunx Wavelet schemes. It has been validated on experimental data, acquired Using synchrotron radiation at the ESRF, Grenoble. The generalization of these reconstruction methods for divergent geometries (2d-3d) is then studied. The main difficulty consists in the direct computation of the radon transform from a divergent data set. Nevertheless, we propose an approximate multiresolution algorithm that makes full use of the particular acquisition geometry in 3D RA and of the special Choice of non separable wavelets. Simulations on mathematical phantoms allowed validating our algorithm and fixing its conditions of Use. Lastly, these multiresolution techniques were successfully applied to 3D rotational angiography. The feasibility of our Reconstruction method w as shown on clinical data, acquired at the Neurological hospital in Lyon. A fast low-resolution reconstruction of the 3D arterial vessels with the progressive addition of details in a Region of interest w as demonstrated
Filali, Wassim. "Détection temps réel de postures humaines par fusion d'images 3D." Toulouse 3, 2014. http://thesesups.ups-tlse.fr/3088/.
Full textThis thesis is based on a computer vision research project. It is a project that allows smart cameras to understand the posture of a person. It allows to know if the person is alright or if it is in a critical situation or in danger. The cameras should not be connected to a computer but embed all the intelligence in the camera itself. This work is based on the recent technologies like the Kinect sensor of the game console. This sensor is a depth sensor, which means that the camera can estimate the distance to every point in the scene. Our contribution consists on combining multiple of these cameras to have a better posture reconstruction of the person. We have created a dataset of images to teach the program how to recognize postures. We have adjusted the right parameters and compared our program to the one of the Kinect
Boulinguez, David. "Reconnaissance 2D et 3D d'objets sous-marins enfouis." Brest, 2000. http://www.theses.fr/2000BRES2019.
Full textNicolas, Yannick. "Construction de représentations 3D compactes de séquences vidéo pour la navigation à distance dans des scènes tridimensionnelles." Rennes 1, 2002. http://www.theses.fr/2002REN10016.
Full textLozes, François. "Traitements d'images sur surfaces et variétés avec mise en application au patrimoine culturel 3D." Caen, 2015. https://hal.archives-ouvertes.fr/tel-01131419.
Full textIn this thesis, we are interesting in transposing and solving PDEs and variational problems on general surfaces or manifolds. PDEs and variational methods are one of the most important tools widely used for modeling and solving inverse problems, e. G. , in image processing and computer vision. Recently, many of these methods were extended to non-local forms. However, most of the research works on local or non-local processing focus only on image processing on Euclidean spaces. In this thesis, we propose an approach to transpose and to solve PDEs on surfaces and point clouds. This latter approach is based on the representation of point clouds by weighted graphs and the framework of Partial differential Equations (PdEs). This approach requires no pre-processing of point clouds, and allow to extend and to adapt non-local PDEs by changing only the topology of the graph
Lei, Ming. "Imagerie 3D d'impédance bioélectrique : problème direct, problème inverse : détermination des lignes de courant en 3D et application de la méthode de la matrice de sensibilité pour la reconstruction d'une image en 3D dans un volume conducteur inhomogène sphérique." Toulouse, INPT, 1995. http://www.theses.fr/1995INPT010H.
Full textLebrun, Justine. "Appariement inexact de graphes appliqué à la recherche d'images et d'objets 3D." Thesis, Cergy-Pontoise, 2011. http://www.theses.fr/2011CERG0510/document.
Full textMany type of documents can be modeled by a graph representation. In this thesis, we focus on theuse of graph for research in multimedia databases.We begin by presenting the theory of graphs and aroundan overview of methods that have been proposed for matching.Then, we are particularly interested in their use for recognitionforms and multimedia indexing.In order to respond in the most generic possible differentresearch problems, we propose to work within the framework of kernel functions.This framework allows to separate the problems related to the nature of the documentsthose introduced by the different types of research. Thus, all ourenergy is devoted to the design of mapping functions,but bearing in mind that they must meet a numbermathematical properties. In this context, we propose newsolutions that better meet the specificgraphs from primitive and visual descriptors. Wealso present algorithms to quickly assessthese functions. Finally, wepresent experiments that highlight thesedifferent characteristics and experiences that showadvantages of our models with respect to the literature
Grasset-Simon, Carine. "Définition et étude des pyramides généralisées ηD : application pour la segmentation multi-échelle d'images 3D." Poitiers, 2006. http://www.theses.fr/2006POIT2353.
Full textIn this work, we are interested in the hierarchical geometrical modeling with a topological basis. We propose the definition of a generic model in any dimension, and we show a possible application in multi-level segmentation of 3D images. In the first part of this work, we define and study the nD generalized pyramids. This is a generic hierarchical topological model which represents all the cells of a subdivision as well as the adjacency and incidence relations existing between these cells. We propose and compare three possible representations of these pyramids. In order to retrieve the information which corresponds to a cell, we define the notion of generalized orbit. This notion extends the notion of receptive field. We also define a local modification operation of a pyramid level allowing to preserve the model coherence by propagating the modifications at the upper levels. In the second part of this work, we show how to use this model in the case of a multi-level segmentation of 3D images. We define the properties which have to be respected by the pyramid, and we give the algorithms that allow to construct such a pyramid. Then, we show how to use the generalized orbits in order to retrieve the voxels or the inter-voxel elements which compose a region or its boundary. Finally we define an operation allowing to locally modify the segmentation criterion of a region set. This operation is based on the operation defined in the first part in order to preserve the coherence constraints
Paquit, Vincent. "Imagerie multispectrale et modélisation 3D pour l'estimation quantitative des vaisseaux sanguins sous-cutanés." Dijon, 2008. http://www.theses.fr/2008DIJOS055.
Full textCurrent methods of venipuncture and peripheral intravenous (IV) catheterization procedures are performed ‘blind’ in that the clinician cannot see the vein but has to rely on feel, which may lead to missing the vein completely, causing tissue damage, pain and anxiety for the patient, especially with small children, older patients and those in a dehydrated state. We present in this dissertation experimental and theoretical methods to (1) Estimate relative vein location in different types of subcutaneous tissue and various skin colors, done by imaging the skin using a discrete set of illuminants ranging from visible to near infrared and a structured light pattern for active 3D triangulation and skin surface shape measurements. (2) Quantitatively optimize the measurement of vein depth in the tissue, using multispectral imaging, three dimensional Monte Carlo photon transport modeling, and inverse imaging with pattern recognition on phantom limbs and human subjects. (3) Develop imaging software enhancements utilizing three-dimensional mapping to help visualize the estimation of an optimal catheter entry point into a vein. The long term goal is to create an easy-to-use, inexpensive portable vein imaging device that will allow vein visualization for IV access and also assessment of IV catheter integrity in order to verify correct IV catheter placement
Bosc, Marcel. "Contribution à la détection de changements dans des séquences IRM 3D multimodales." Phd thesis, Université Louis Pasteur - Strasbourg I, 2003. http://tel.archives-ouvertes.fr/tel-00005163.
Full textSebbahi, Ali. "Segmentation 2D et 3D par modèles déformables en imagerie cardio-vasculaire." Paris 12, 1995. http://www.theses.fr/1995PA120040.
Full textYANG, RUI LIANG. "Traitement numerique des angiogrammes : videodensitometrie et reconstruction d'images en 3d a partir de deux projections orthogonales." Paris, CNAM, 1989. http://www.theses.fr/1989CNAM0105.
Full textGuislain, Maximilien. "Traitement joint de nuage de points et d'images pour l'analyse et la visualisation des formes 3D." Thesis, Lyon, 2017. http://www.theses.fr/2017LYSE1219/document.
Full textRecent years saw a rapid development of city digitization technologies. Acquisition campaigns covering entire cities are now performed using LiDAR (Light Detection And Ranging) scanners embedded aboard mobile vehicles. These acquisition campaigns yield point clouds, composed of millions of points, representing the buildings and the streets, and may also contain a set of images of the scene. The subject developed here is the improvement of the point cloud using the information contained in the camera images. This thesis introduces several contributions to this joint improvement. The position and orientation of acquired images are usually estimated using devices embedded with the LiDAR scanner, even if this information is inaccurate. To obtain the precise registration of an image on a point cloud, we propose a two-step algorithm which uses both Mutual Information and Histograms of Oriented Gradients. The proposed method yields an accurate camera pose, even when the initial estimations are far from the real position and orientation. Once the images have been correctly registered, it is possible to use them to color each point of the cloud while using the variability of the point of view. This is done by minimizing an energy considering the different colors associated with a point and the potential colors of its neighbors. Illumination changes can also change the color assigned to a point. Notably, this color can be affected by cast shadows. These cast shadows are changing with the sun position, it is therefore necessary to detect and correct them. We propose a new method that analyzes the joint variation of the reflectance value obtained by the LiDAR and the color of the points. By detecting enough interfaces between shadow and light, we can characterize the luminance of the scene and to remove the cast shadows. The last point developed in this thesis is the densification of a point cloud. Indeed, the local density of a point cloud varies and is sometimes insufficient in certain areas. We propose a directly applicable approach to increase the density of a point cloud using multiple images
Ardon, Roberto. "Extraction de surfaces dans des images 3D et chemins minimaux : applications en imagerie médicale." Paris 9, 2005. https://portail.bu.dauphine.fr/fileviewer/index.php?doc=2005PA090010.
Full textDeveau, Matthieu. "Utilisation conjointe de données image et laser pour la segmentation et la modélisation 3D." Paris 5, 2006. http://www.theses.fr/2006PA05S001.
Full textThis thesis deals with combining a digital image with laser data for complex scenes 3D modeling automation. Image and laser data are acquired from the same point of view, with a greater image resolution than the laser one. This work is structured around three main topics pose estimation, segmentation and modeling. Pose estimation is based both on feature points matching, between the digital image and the laser intensity image, and on linear feature extraction. Data segmentation into geometric features is done through a hierarchical segmentation scheme, where image and laser data are combined. 3D modeling automation is studied through this hierarchical scheme. A tool for semi-automated modeling is also derived from the hierarchical segmentation architecture. In the modeling step, we have focused on automatic modeling of cylinders with free-form profiles. The description is then very general, with planes, freeform profile cylinders, revolution objects, and meshes on complex parts