Rozprawy doktorskie na temat „Traitement d'image vectorielle”
Utwórz poprawne odniesienie w stylach APA, MLA, Chicago, Harvard i wielu innych
Sprawdź 24 najlepszych rozpraw doktorskich naukowych na temat „Traitement d'image vectorielle”.
Przycisk „Dodaj do bibliografii” jest dostępny obok każdej pracy w bibliografii. Użyj go – a my automatycznie utworzymy odniesienie bibliograficzne do wybranej pracy w stylu cytowania, którego potrzebujesz: APA, MLA, Harvard, Chicago, Vancouver itp.
Możesz również pobrać pełny tekst publikacji naukowej w formacie „.pdf” i przeczytać adnotację do pracy online, jeśli odpowiednie parametry są dostępne w metadanych.
Przeglądaj rozprawy doktorskie z różnych dziedzin i twórz odpowiednie bibliografie.
Even, Melvin. "Animation 2D esquissée assistée par ordinateur". Electronic Thesis or Diss., Bordeaux, 2024. http://www.theses.fr/2024BORD0457.
Pełny tekst źródłaTraditional 2D animation is a time-intensive process requiring a high level of expertise, as animators must draw thousands of individual frames by hand to create a complete animation sequence. With the advent of computer animation, artists now have the option to use 3D animation or 2D puppet-style techniques, which eliminate the need to manually draw each frame but often sacrifice the unique workflows and artistic style characteristic of traditional 2D animation. Alternatively, digital tools that emulate traditional 2D animation streamline tasks like drawing and coloring, but still require animators to draw each frame by hand. In academia, many computer-assisted methods have been developed to automate the inbetweening of clean line drawings, in which intermediate frames are automatically generated from input key drawings. However, these methods tend to be constrained by rigid workflows and limited artistic control, largely due to the challenges of stroke matching and interpolation. In this thesis, we take a novel approach to address these limitations by focusing on an earlier stage of animation using rough drawings (i.e., sketches). Our key innovation is to recast the matching and interpolation problems using transient embeddings, which consist in groups of strokes that exist temporarily across keyframes. A transient embedding carries strokes between keyframes both forward and backward in time through a sequence of transformed lattices. Our system generates rough inbetweens in real-time allowing artists to preview and edit their animation using tools that offer precise artistic control over the dynamics of the animation. This ensures smooth continuity of motion, even when complex topological changes are introduced and enables non-linear exploration of movements. We demonstrate these capabilities on state-of-the-art 2D animation examples. Another notoriously difficult task is the representation 3D motion and depth through 2D animated drawings. Artists must pay particular attention to occlusions and how they evolve through time, a tedious process. Computer-assisted inbetweening methods such as cut-out animation tools allow for such occlusions to be handled beforehand using a 2D rig, at the expense of flexibility and artistic expression. We also address occlusion handling without sacrificing non-linear control. Our contribution in that matter is two-fold: a fast method to compute 2D masks from rough drawings with a semi-automatic dynamic layout system for occlusions between drawing parts; and an artist-friendly method to both automatically and manually control the dynamic visibility of strokes for self-occlusions. Our system helps artists produce convincing 3D-like 2D animations, including head turns, foreshortening effects, out-of-plane rotations, overlapping volumes and even transparency
Chanussot, Jocelyn. "Approches vectorielles ou marginales pour le traitement d'images multi-composantes". Chambéry, 1998. http://www.theses.fr/1998CHAMS025.
Pełny tekst źródłaChevaillier, Béatrice. "Analyse de données d'IRM fonctionnelle rénale par quantification vectorielle". Electronic Thesis or Diss., Metz, 2010. http://www.theses.fr/2010METZ005S.
Pełny tekst źródłaDynamic-Contrast-Enhanced Magnetic Resonance Imaging has a great potential for renal function assessment but has to be evaluated on a large scale before its clinical application. Registration of image sequences and segmentation of internal renal structures is mandatory in order to exploit acquisitions. We propose a reliable and user-friendly tool to partially automate these two operations. Statistical registration methods based on mutual information are tested on real data. Segmentation of cortex, medulla and cavities is performed using time-intensity curves of renal voxels in a two step process. Classifiers are first built with pixels of the slice that contains the largest proportion of renal tissue : two vector quantization algorithms, namely the K-means and the Growing Neural Gas with targeting, are used here. These classifiers are first tested on synthetic data. For real data, as no ground truth is available for result evaluation, a manual anatomical segmentation is considered as a reference. Some discrepancy criteria like overlap, extra pixels and similarity index are computed between this segmentation and functional one. The same criteria are also evaluated between the referencee and another manual segmentation. Results are comparable for the two types of comparisons. Voxels of other slices are then sorted with the optimal classifier. Generalization theory allows to bound classification error for this extension. The main advantages of functional methods are the following : considerable time-saving, easy manual intervention, good robustness and reproductibility
Lee, Hyun-Soo. "Étude comparative de la classification de textures fondée sur la représentation vectorielle". Le Havre, 1992. http://www.theses.fr/1992LEHA0002.
Pełny tekst źródłaChevaillier, Béatrice. "Analyse de données d'IRM fonctionnelle rénale par quantification vectorielle". Phd thesis, Université de Metz, 2010. http://tel.archives-ouvertes.fr/tel-00557235.
Pełny tekst źródłaRafaa, Az-Eddine. "Représentation multirésolution et compression d'images : ondelettes et codage scalaire et vectoriel". Metz, 1994. http://docnum.univ-lorraine.fr/public/UPV-M/Theses/1994/Rafaa.Az_Eddine.SMZ9447.pdf.
Pełny tekst źródłaWavelet pyramidal is a new form of image representation. Its coding scheme allows a new method of image data compression. This technical may well supplant the well established block-transform methods which have been state of art for these last years. In a suitable purpose, this work is achieved in three parts. The first part provides theorical and historical synthesis of multiresolution representation which converge into Wavelets pyramidal representation of images. The second part investigate scalar point of view of different pyramid components : statistical characteristics, coding scheme with scalar quantization or with dead-zone, related to optimal bit allocation whether analytical or algorithmic. Finally, the third part deals with vector coding scheme. The pyramid is then dispatched into elementary entities which are performed on a hierarchical vector tree as which we applied an hybrid coding based on QS and QV. This structure allows local directivity and frequency activity description in image, and then an introduction to the notion of activity classification. To extend the validity of the elaborated encoder, extensions to colours images and images sequences was presented too
Akrout, Nabil. "Contribution à la compression d'images par quantification vectorielle : algorithmes et circuit intégré spécifique". Lyon, INSA, 1995. http://www.theses.fr/1995ISAL0017.
Pełny tekst źródłaRecently, Vector Quantization (VQ) has received considerable attention and become an effective tool for image compression. It provides High compression ratio and simple decoding process. However, studies on practical implementation of VQ have revealed some major difficulties such as edge integrity and code book design efficiency. After doing the state-of-the-art in the field of Vector Quantization, we focus on: - Iterative and non-iterative code book generation algorithms. The main idea of non-iterative algorithms consists of creating progressively the code words, during only one scanning of the training set. At the beginning, the code book is initialized by the first vector found in the training set, then each input vector is mapped into the nearest neighbor codeword, which minimizes the distortion error. This error is then compared with per-defined thresholds. The performance of iterative and non-iterative code book generation algorithms are compared: the code books generated by non-iterative algorithms require less than 2 percent of the time required by iterative algorithms. - To propose a new procedure for image compression as an improvement on vector quantization of sub-bands. Codewords having vertical and horizontal shape will be used to vector quantize the high-frequency sub-images, obtained from a multiresolution analysis scheme. The codewords shapes take into account the orientation and resolution of each subband details in order to -preserve edges at low bit rates. Their sizes are defined according to the correlation distances in each subband for the horizontal and vertical directions. - The intensive computational demands of vector quantization (V. Q. ) for important applications in speech and image compression have motivated the need for dedicated processors with very high throughput capabilities. Bit-serial systolic architectures offer one of the most promising approaches for fulfilling the demanding V. Q. Speed requirements in many applications. We propose a novel family of architectural techniques which offer efficient computation of Manhattan distance measures for nearest neighbor code book searching. Manhattan distance requires much less computation and VLSI chip area, because there is no multiplier. Compared to Euclidean distance. This gave rise to the idea of implementing Manhattan distance directly in hardware for real-time image coding. Very high V. Q. Throughput can be achieved by a massive parallelism. Therefore, it requires an important VLSI chip area. To avoid this difficulty, we carry out a "bit-serial" pipelined processing for nearest neighbor code book searching. This architecture is more suitable for real-time coding. Several alternative configurations allow reasonable tradeoffs between speed and VLSI chip area required
Alshatti, Wael. "Approches vectorielles du filtrage et de la détection de contours dans des images multi-spectrales". Chambéry, 1994. http://www.theses.fr/1994CHAMS022.
Pełny tekst źródłaFurlan, Gilbert. "Contribution à l'étude et au développement d'algorithmes de traitement du signal en compression de données et d'images". Nice, 1990. http://www.theses.fr/1990NICE4433.
Pełny tekst źródłaMoravie, Philippe. "Parallélisation d'une méthode de compression d'images : transformée en ondelettes, quantification vectorielle et codage d'Huffman". Toulouse, INPT, 1997. http://www.theses.fr/1997INPT123H.
Pełny tekst źródłaBatard, Thomas. "Géométrie différentielle des fibrés vectoriels et algèbres de Clifford appliquées au traitement d'images multicanaux". Phd thesis, Université de La Rochelle, 2009. http://tel.archives-ouvertes.fr/tel-00684250.
Pełny tekst źródłaLennon, Marc. "Methodes d'analyse d'images hyperspectrales : exploitation du capteur aéroporté CASI pour des applications de cartographie agro-environnementale en Bretagne". Rennes 1, 2002. http://www.theses.fr/2002REN10130.
Pełny tekst źródłaPignol, Michel. "Realisation d'un mini-processeur vectoriel travaillant en arithmetique flottante 32 bits, adapte au traitement frequentiel d'images". Clermont-Ferrand 2, 1986. http://www.theses.fr/1986CLF2D181.
Pełny tekst źródłaPignol, Michel. "Réalisation d'un mini-processeur vectoriel travaillant en arithmétique flottante 32 bits, adapté au traitement fréquentiel d'images". Grenoble 2 : ANRT, 1986. http://catalogue.bnf.fr/ark:/12148/cb37600413t.
Pełny tekst źródłaAnwander, Alfred. "Segmentation d'images couleur par un opérateur gradient vectoriel multiéchelle et contour actif : application à la quantification des phases minéralogiques du clinker de cimen". Lyon, INSA, 2001. http://theses.insa-lyon.fr/publication/2001ISAL0062/these.pdf.
Pełny tekst źródłaThe objective of this thesis is the quantification of cement clinker phases by analysis of color images acquired with an optical microscope. The segmentation of the crystals is approached in a double way adapted to industrial constraints. The first introduces an orignal computation of the color gradient for images acquired with a standard mono-CCD color camera. These multispectral images have the particularity of a reduced resolution of color components compared to that of luminance. The adaptation of a multiscale approach and weighting of the components are integrated in a new multiscale color gradient (MCG) operator. Comparative studies with classical techniques in the domain of edge detection, show the contribution of the MCG. The second original model concerns image segmentation by deformable contours of “balloon” type. Our model stabilizes the deformation by normalizing the contour length at each iteration. This allows big deformations of the contour without re-sampling of the contour, which is necessary in the classical approach. The algorithm is faster, more robust and allows an initial contour which is far from the final contour. The software developed in partnership with the company “Lafarge Ciments” integrates the different color image processing techniques. Pre-processing of the images, segmentation by edge detection using the MCG of the microscopic color images and edge closing. The homogeneous regions segmentation in this way are classified using color and texture parameters. The system computes a quantification of the three clinker phases and a statistical analysis of the crystal dimensions. It replaces long and tiring manual measurements and allows a high quality production by more frequent measurements at the exit of the cement kiln
Allanic, Hervé. "Utilisation du modèle vectoriel topologique pour la reconnaissance des formes". Paris 12, 2002. http://www.theses.fr/2002PA120026.
Pełny tekst źródłaThis thesis presents a work on an image understanding system based on semantic networks which aim is to realize data fusion with symbolic primitives extracted from an image. We use a Topological Vectorial Model (MVT) which is broadly used in the Geographic Information Systems. This model can handle semantic informations together with geometrical datas. After looking at the relevance of the MVT for the image understanding area, we have defined the characteristics of a system based on this model. We have then been able to define data and task types, to create system architecture and to validate the approach on two applications
Anwander, Alfred Neyran Bruno Baskurt Atilla. "Segmentation d'images couleur par un opérateur gradient vectoriel multiéchelle et contour actif application à la quantification des phases minéralogiques du clinker de ciment /". Villeurbanne : Doc'INSA, 2005. http://docinsa.insa-lyon.fr/these/pont.php?id=anwander.
Pełny tekst źródłaClienti, Christophe. "Architectures flot de données dédiées au traitement d'images par morphologie mathématique". Phd thesis, École Nationale Supérieure des Mines de Paris, 2009. http://pastel.archives-ouvertes.fr/pastel-00005758.
Pełny tekst źródłaAptoula, Erchan. "Morphological colour image analysis : Applied to content-based image description, annotation and retrieval". Université Louis Pasteur (Strasbourg) (1971-2008), 2008. https://publication-theses.unistra.fr/public/theses_doctorat/2008/APTOULA_Erchan_2008.pdf.
Pełny tekst źródłaThis thesis concentrates mainly on the extension of mathematical morphology to colour images. The resulting morphological operators are then applied to the problem of general purpose content-based colour image retrieval. First the choice of a colour space is discussed and polar spaces of type luminance-saturation-hue are endorsed on the grounds of their intuitiveness, as far as the human visual perception is concerned. The rest of the first part focuses on the sole requirement for the extension of mathematical morphology to multivariate data: a vector ordering. Specifically, due to its multiple positive properties, we concentrate on the lexicographical approach. We further introduce several variations to it, with the purpose of rendering possible a finer tuning of the processing balance among image channels, and between chromatic and achromatic information in the particular of colour images. Their practical interest is illustrated through a comparative study, that is carried out on the properties of several vector orderings. Moreover, considering the periodic nature of hue values, that does not lend itself to scalar ordering, an image specific approach based on multiple reference hues is proposed. This part is concluded with the extension of the hit-or-miss transform to colour images, as an example based on its findings. The second part is dedicated to the application of colour morphology to the problem of content-based image retrieval. To this end, novel morphology based colour and texture descriptors, using the principles developed in the first part, are presented and compared against state-of-the-art approaches with multiple image databases. In particular, multiscale histogram variations are studied as global colour descriptors, whereas the information extracted by granulometry and morphological covariance is combined with the purpose of texture characterisation. Finally a semi-supervised, keyword-based image annotation and retrieval architecture is developed, based on the aforementioned operators, leading to a flexible solution in terms of adaptability to various image collection heterogeneity levels, providing feedback support and capable of semantic queries
Zhao, Yili. "Étude et réalisation d'un système de transmission d'images numériques à bas débit". Nancy 1, 1989. http://www.theses.fr/1989NAN10266.
Pełny tekst źródłaCachier, Pascal. "Recalage non rigide d'images médicales volumiques : contributions aux approches iconiques et géométriques". Phd thesis, Ecole Centrale Paris, 2002. http://tel.archives-ouvertes.fr/tel-00630493.
Pełny tekst źródłaOrzan, Alexandrina. "Images basées sur les contours : représentation, création et manipulation". Grenoble INPG, 2009. http://www.theses.fr/2009INPG0179.
Pełny tekst źródłaThis thesis proposes a novel image primitive—the diffusion curve. This primitive relies on the principle that images can be defined via their discontinuities, and concentrates image features along contours. The diffusion curve can be defined in vector graphics, as well as in raster graphics, to increase user control during the process of art creation. The vectorial diffusion curve primitive augments the expressive powers of vector images by capturing complex spatial appearance behaviours. Diffusion curves represent a simple and easy to manipulate support for complex content representation and edition. In raster images, diffusion curves define a higher level structural organization of the pixel image. This structure is used to create simplified or exaggerated representations of photographs in a way consistent with the original image content. Finally, a fully automatic vectorization method is presented, that converts raster diffusion curve to vector diffusion curve
Alata, Olivier. "Contributions à la description de signaux, d'images et de volumes par l'approche probabiliste et statistique". Habilitation à diriger des recherches, Université de Poitiers, 2010. http://tel.archives-ouvertes.fr/tel-00573224.
Pełny tekst źródłaTang, Hui. "Méthodologies de visualisation 3D en imagerie médicale". Phd thesis, Université Rennes 1, 2008. http://tel.archives-ouvertes.fr/tel-00355629.
Pełny tekst źródłaL'uroscan fournit 3 à 4 volumes présentant une information complémentaire sur l'anatomie rénale. La première étape consiste à mettre en correspondance ces différents volumes par une technique de recalage rigide du volume rénal basée sur la maximisation locale de l'information mutuelle.
L'idée principale de ce mémoire de Thèse est de proposer une visualisation de l'anatomie rénale directement à partir de ces données fusionnées. Pour cela, une technique de classification statistique des données basée sur une modélisation de la distribution des valeurs par un mélange de Gaussiennes incluant une information spatiale a été développée. Différentes techniques de visualisation 3D ont ensuite été adaptées à la représentation de cette information et comparées entre-elles.
Les techniques de représentation de surfaces peuvent être accélérées par des procédures de simplifications de maillages. Dans ce cadre, nous avons proposé deux métriques de description de la surface basées sur les moments géométriques et pouvant être incluses dans une telle procédure.
Ces différentes solutions, même si elles ont été développées dans le cadre de la représentation des structures anatomiques rénale, sont suffisamment génériques pour être utilisées ou adaptées à d'autres organes ou à d'autres applications médicales.