Dissertations / Theses on the topic 'Protocoles d'Évaluation'
Create a spot-on reference in APA, MLA, Chicago, Harvard, and other styles
Consult the top 18 dissertations / theses for your research on the topic 'Protocoles d'Évaluation.'
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.
Raynal, Frédéric. "Etudes d'outils pour la dissimulation d'information : approches fractales, protocoles d'évaluation et protocoles cryptographiques." Paris 11, 2002. http://www.theses.fr/2002PA112020.
Full textData hiding both relies on signal processing and cryptography, which constitute the main two parts of this thesis. Some watermarking schemes are based on fractal compression algorithms, i. E. On the resolution of the inverse problem for Iterated Function Systems (IFS). It consists in finding an IFS resulting in an attractor as near as possible to a given pattern. To this end, we propose an evolutionary algorithm-based stochastic optimization technique, developing the "Parisian Evolution" paradigm in which the solution is modeled as a subset of an evolutionary population of functions. Although an improvement over the classical EA approach, this algorithm remains computationally expensive. Next, we introduce the principles of a watermarking scheme based on multifractal spectra analysis. In spite of the interesting robustness of the regularity information, this approach did not provide both the discriminance and reliability required for watermarking. The third part of the document presents an evaluation method for watermarking schemes and the Stirmark benchmarking software, which uses classical attacks on fixed image and audio files, and also provides new tests like key space issues, multiple marking or false alarm rate. Several important notions of cryptography and their applicability to data hiding are investigated. Although these are strictly speaking different problems, we showed that several concepts could be integrated directly into the diagram (zero-knowledge proof or secret sharing) or into the protocol (asymmetrical schemes or digital signature). We also address new cryptographic problems like asymmetric encryption with multiple keys and soft hash functions. Special attention is given to covert channels in computer networks. After analyzing existing solutions, we show that the "Applications" layer of the OSI model contains such a channel in the SSH protocol
Claudet, Joachim. "Aires marines protégées et récifs artificiels : méthodes d'évaluation, protocoles expérimentaux et indicateurs." Perpignan, 2006. http://www.theses.fr/2006PERP0736.
Full textMarine Protected Areas’ (MPAs) and Artificial Reef’s (ARs) management requires complex assessment and monitoring programmes, dealing with different sources of variability. We studied and developed experimental designs and analysis methods suited for the establishment of a monitoring of MPAs and ARs. This methodology is developed from existing data sets in the Northwestern Mediterranean. We build multi-criteria indicators allowing a statistically testable diagnosis of the impact of MPAs and ARs on reef fish assemblages. Using ecological performance indicators permits to monitor and to give an image of the assessed system to managers. It was possible to show the global response of the fish assemblages to the protection by a MPA. This response was evidenced by increases in abundance, species richness or diversity, gradually through space, time and among various taxonomic groups or fish individual sizes. Large fishes reacted faster to protection and shallow habitats were more sensitive to the existence of a MPA. Our results can be useful for the implementation of new MPAs or for the immersions of ARs and for the development of their management plans. Key-words : Marines Protected Areas, Artificial Reefs, Impact Assessment, Temperate Fish, Indicators, Multivariate, Habitat, Monitoring, Statistical Power, Northwestern Mediterranean, Management
Modrzejewski, Richard. "Recalage déformable, jeux de données et protocoles d'évaluation pour la chirurgie mini-invasive abdominale augmentée." Thesis, Université Clermont Auvergne (2017-2020), 2020. http://www.theses.fr/2020CLFAC044.
Full textThis thesis deals with deformable registration techniques of preoperative data to the intra-operative sceneas an indispensable step in the realisation of augmented reality for abdominal surgery. Such techniques arethus discussed as well as evaluation methodologies associated with them. Two contexts are considered : theregistration for computer-assisted laparoscopic surgery and the postural registration of the patient on theoperating table. For these two contexts, the needs to be met by the registration algorithms considered arediscussed, as well as the main limitations of the existing solutions. Algorithms developped during this thesis,allowing to meet these needs are thus proposed and discussed. Special attention is given to their evaluation.Different datasets allowing a quantitative evaluation of the accuracy of the registration algorithms, also realizedduring this thesis, and made public, are also discussed. Such data are extremely important because they respondto a lack of evaluation data needed in order to evaluate the registration error in a quantitative way, and thus tocompare the different algorithms. The modeling of the illumination of the laparoscopic scene, allowing one toextract strong constraints between the data to be registered and the surface of the observed organ, and thus tobe used to constrain these registration problems, is also discussed. This manuscript has seven parts. The firstdeals with the context surrounding this thesis. Minimally invasive surgery is presented as well as various generalcomputer vision problems which, when applied to the medical context, allow the definition of computer-assistedsurgery. The second part deals with the prerequisites for reading the thesis. The pre-processing of pre-operativeand per-operative data, before their use by the presented registration algorithms, is thus discussed. The thirdpart corresponds to the registration of hepatic data in laparoscopy, and the evaluation associated with thisproblems. The fourth part deals with the problem of postural registration. The fifth part proposes a modellingof the lighting in laparoscopy which can be used to obtain strong constraints between the observed surfaceand the laparoscopic images. The sixth part proposes a use of the light models discussed in the previous partin order to refine and densify reconstructions of the laparoscopic scene. Finally, the seventh and last partcorresponds to our conclusions regarding the issues addressed during this thesis, and to future work
Mendoza, Herrera Andrés Dario. "La famille prend la parole : évaluation de l'implantation d'un protocole d'évaluation familiale (PEF) dans l'intervention clinique." Master's thesis, Université Laval, 2017. http://hdl.handle.net/20.500.11794/27949.
Full textBerthet, Alexandre. "Deep learning methods and advancements in digital image forensics." Electronic Thesis or Diss., Sorbonne université, 2022. http://www.theses.fr/2022SORUS252.
Full textThe volume of digital visual data is increasing dramatically year after year. At the same time, image editing has become easier and more precise. Malicious modifications are therefore more accessible. Image forensics provides solutions to ensure the authenticity of digital visual data. Recognition of the source camera and detection of falsified images are among the main tasks. At first, the solutions were classical methods based on the artifacts produced during the creation of a digital image. Then, as in other areas of image processing, the methods moved to deep learning. First, we present a state-of-the-art survey of deep learning methods for image forensics. Our state-of-the-art survey highlights the need to apply pre-processing modules to extract artifacts hidden by image content. We also highlight the problems concerning image recognition evaluation protocols. Furthermore, we address counter-forensics and present compression based on artificial intelligence, which could be considered as an attack. In a second step, this thesis details three progressive evaluation protocols that address camera recognition problems. The final protocol, which is more reliable and reproducible, highlights the impossibility of state-of-the-art methods to recognize cameras in a challenging context. In a third step, we study the impact of compression based on artificial intelligence on two tasks analyzing compression artifacts: tamper detection and social network recognition. The performances obtained show on the one hand that this compression must be taken into account as an attack, but that it leads to a more important decrease than other manipulations for an equivalent image degradation
Giovanni, Antoine. "Contribution à l'élaboration d'un protocole multiparamétrique d'évaluation objective des dysphonies." Aix-Marseille 2, 1998. http://theses.univ-amu.fr.lama.univ-amu.fr/1998AIX2668U.pdf.
Full textBoudra, Abdelhamid. "Protocole d'évaluation de la contamination mycotoxique. Application à la noix et à la figue." Toulouse 3, 1994. http://www.theses.fr/1994TOU30066.
Full textDe, Avellar Dantas Ana Regina. "Protocole d'évaluation de la fatigabilité des muscles du dos au cours de la grossesse." Compiègne, 2004. http://www.theses.fr/2004COMP1518.
Full textBack pain is a common problem amongst pregnant women. We have developed a protocol to evaluate the fatigability of trunk extensor muscles (spinalis thoracis, longissimus lumborum) by means of surface electromyogram spectral parameters; this proctocol also includes measurements of postural equilibrium, ligamentary laxity and lordosis during pregnancy. We have first tested it with 3 groups of non-pregnant women: a control group tested in the moming and the evening, a reproducibility group and a simulation group carrying a « false baby» to induce supplementary work on the back muscles. Next, the protocol has been applied to pregnant women
Madrigrano, Gina Danielle. "Vers la validation d'un protocole d'évaluation pour les adolescents ayant commis des crimes sexuels, entrevue et questionnaires." Thesis, National Library of Canada = Bibliothèque nationale du Canada, 2000. http://www.collectionscanada.ca/obj/s4/f2/dsk2/ftp02/NQ52174.pdf.
Full textSelmi, Wissal. "Services écosystémiques rendus par la végétation urbaine : application d'approches d'évaluation à la ville de Strasbourg." Thesis, Strasbourg, 2014. http://www.theses.fr/2014STRAH013/document.
Full textThis research work provides two assessment tools based on both ecocentric and anthropogenic approaches. We argued that these approaches are complementary and they lead not only to understand ecological functions of urban green spaces but also to quantify ecosystem services provided to society. Based on bottom up approach, two urban habitats were assessed : lawns and urban forest. Due to the lack of knowledge about ecological functions of urban lawns, it was required to implement a monitoring protocol that helps to provide a baseline and measure the changes of flora composition and structure across urban green spaces. Although monitoring protocol had some limitations, it intended to highlight the response of lawn flora to environmental patterns and to particular human activities such as management techniques and trampling. Urban forest was assessed by quantifying it structure and some ecosystem services and desservices using i-Tree model Eco. Based on biophysical indicators, the model quantify the total carbon stored and the annually carbon sequestered, the annually amount of pollution removal, and the annually amount of biogenic emissions by trees. Although some uncertainty remains about the application of this model, it was shown that urban trees improve local air quality. However, to alleviate air pollution within urban area, planting and managing trees should be associated with an integrative planning strategy that takes into account other factors. Our study also incorporates operational items, so we have tried to provide some guidance to planners and green spaces managers with reference to our assessment results
Pineau, Lionel. "Intérêt des biofilms bactériens dans la mise au point d'un protocole d'évaluation des machines à laver et à désinfecter les endoscopes." Toulouse, INPT, 1996. http://www.theses.fr/1996INPT017A.
Full textCalarasanu, Stefania Ana. "Improvement of a text detection chain and the proposition of a new evaluation protocol for text detection algorithms." Thesis, Paris 6, 2015. http://www.theses.fr/2015PA066524/document.
Full textThe growing number of text detection approaches proposed in the literature requires a rigorous performance evaluation and ranking. An evaluation protocol relies on three elements: a reliable text reference, a matching strategy and finally a set of metrics. The few existing evaluation protocols often lack accuracy either due to inconsistent matching or due to unrepresentative metrics. In this thesis we propose a new evaluation protocol that tackles most of the drawbacks faced by currently used evaluation methods. This work is focused on three main contributions: firstly, we introduce a complex text reference representation that does not constrain text detectors to adopt a specific detection granularity level or annotation representation; secondly, we propose a set of matching rules capable of evaluating any type of scenario that can occur between a text reference and a detection; and finally we show how we can analyze a set of detection results, not only through a set of metrics, but also through an intuitive visual representation. A frequent challenge for many Text Understanding Systems is to tackle the variety of text characteristics in born-digital and natural scene images for which current OCRs are not well adapted. For example, texts in perspective are frequently present in real-word images because the camera capture angle is not normal to the plane containing the text regions. Despite the ability of some detectors to accurately localize such text objects, the recognition stage fails most of the time. In this thesis we also propose a rectification procedure capable of correcting highly distorted texts evaluated on a very challenging dataset
Aura, Karine. "Protocole d'évaluation du langage fondé sur le traitement de fonctions prosodiques : étude exploratoire de deux patients atteints de gliomes de bas grade en contexte péri-opératoire." Phd thesis, Université Toulouse le Mirail - Toulouse II, 2012. http://tel.archives-ouvertes.fr/tel-00798667.
Full textCalarasanu, Stefania Ana. "Improvement of a text detection chain and the proposition of a new evaluation protocol for text detection algorithms." Electronic Thesis or Diss., Paris 6, 2015. http://www.theses.fr/2015PA066524.
Full textThe growing number of text detection approaches proposed in the literature requires a rigorous performance evaluation and ranking. An evaluation protocol relies on three elements: a reliable text reference, a matching strategy and finally a set of metrics. The few existing evaluation protocols often lack accuracy either due to inconsistent matching or due to unrepresentative metrics. In this thesis we propose a new evaluation protocol that tackles most of the drawbacks faced by currently used evaluation methods. This work is focused on three main contributions: firstly, we introduce a complex text reference representation that does not constrain text detectors to adopt a specific detection granularity level or annotation representation; secondly, we propose a set of matching rules capable of evaluating any type of scenario that can occur between a text reference and a detection; and finally we show how we can analyze a set of detection results, not only through a set of metrics, but also through an intuitive visual representation. A frequent challenge for many Text Understanding Systems is to tackle the variety of text characteristics in born-digital and natural scene images for which current OCRs are not well adapted. For example, texts in perspective are frequently present in real-word images because the camera capture angle is not normal to the plane containing the text regions. Despite the ability of some detectors to accurately localize such text objects, the recognition stage fails most of the time. In this thesis we also propose a rectification procedure capable of correcting highly distorted texts evaluated on a very challenging dataset
Chambon, Sylvie. "Mise en correspondance stéréoscopique d'images couleur en présence d'occultations." Phd thesis, Université Paul Sabatier - Toulouse III, 2005. http://tel.archives-ouvertes.fr/tel-00011437.
Full textDans un premier chapitre, nous établissons un état de l'art des méthodes de mise en correspondance de pixels. Nous donnons un modèle générique des méthodes s'appuyant sur la définition d'éléments constituants. Nous distinguons alors quatre catégories de méthodes : les méthodes locales, les méthodes globales, les méthodes mixtes et les méthodes à multiples passages.
Le second chapitre aborde le problème de l'évaluation des méthodes de mise en correspondance de pixels. Après avoir donné un état de l'art des protocoles existants, nous proposons un protocole d'évaluation et de comparaison qui prend en compte des images avec vérité terrain et qui distingue différentes zones d'occultations.
Dans le troisième chapitre, nous proposons une taxonomie des mesures de corrélation regroupées en cinq familles : les mesures de corrélation croisée, les mesures utilisant des outils de statistiques classiques, les mesures utilisant les dérivées des images, les mesures s'appuyant sur des outils des statistiques non paramétriques et les mesures exploitant des outils des statistiques robustes. Parmi cette dernière famille, nous proposons dix-sept mesures. Les résultats obtenus avec notre protocole montrent que ces mesures obtiennent les meilleurs résultats dans les zones d'occultations.
Le quatrième chapitre concerne la généralisation à la couleur des méthodes de mise en correspondance à base de corrélation. Après avoir présenté les systèmes de représentation de la couleur que nous testons, nous abordons la généralisation des méthodes à base de corrélation en passant par l'adaptation des mesures de corrélation à la couleur. Nous proposons trois méthodes différentes : fusion des résultats sur chaque composante, utilisation d'une analyse en composante principale et utilisation d'une mesure de corrélation couleur. Les résultats obtenus avec notre protocole mettent en évidence la meilleure méthode qui consiste à fusionner les scores de corrélation.
Dans le dernier chapitre, pour prendre en compte les occultations, nous proposons des méthodes hybrides qui s'appuient sur l'utilisation de deux mesures de corrélation : une mesure classique dans les zones sans occultation et une mesure robuste dans les zones d'occultations. Nous distinguons quatre types de méthodes à base de détection de contours, de corrélation pondérée, de post-détection des occultations et de fusion de cartes de disparités. Les résultats obtenus avec notre protocole montrent que la méthode la plus performante consiste à fusionner deux cartes de disparités.
Michaud, Pierre-Yohan. "Distorsions des systèmes de reproduction musicale : Protocole de caractérisation perceptive." Phd thesis, Aix-Marseille Université, 2012. http://tel.archives-ouvertes.fr/tel-00806288.
Full textChassat, Fabrice. "Étude et réalisation d'un système de localisation en six dimensions pour la chirurgie assistée par ordinateur." Phd thesis, Grenoble 1, 2001. http://tel.archives-ouvertes.fr/tel-00005276.
Full textLéchelle, William. "Protocoles d'évaluation pour l'extraction d'information libre." Thèse, 2019. http://hdl.handle.net/1866/22659.
Full text