A recuperação de imagens baseada em conteúdo (Content-based Image Retrieval - CBIR) embasa-se sobre dois aspectos primordiais, um extrator de características o qual deve prover as características intrínsecas mais significativas dos dados e uma função de distância a qual quantifica a similaridade entre tais dados. O grande desafio é justamente como alcançar a melhor integração entre estes dois aspectos chaves com intuito de obter maior precisão nas consultas por similaridade. Apesar de inúmeros esforços serem continuamente despendidos para o desenvolvimento de novas técnicas de extração de características, muito pouca atenção tem sido direcionada à importância de uma adequada associação entre a função de distância e os extratores de características. A presente Dissertação de Mestrado foi concebida com o intuito de preencher esta lacuna. Para tal, foi realizada a análise do comportamento de diferentes funções de distância com relação a tipos distintos de vetores de características. Os três principais tipos de características intrínsecas às imagens foram analisados, com respeito a distribuição de cores, textura e forma. Além disso, foram propostas duas novas técnicas para realização de seleção de características com o desígnio de obter melhorias em relação à precisão das consultas por similaridade. A primeira técnica emprega regras de associação estatísticas e alcançou um ganho de até 38% na precisão, enquanto que a segunda técnica utilizando a entropia de Shannon alcançou um ganho de aproximadamente 71% ao mesmo tempo em que reduz significantemente a dimensionalidade dos vetores de características. O presente trabalho também demonstra que uma adequada utilização das funções de distância melhora efetivamente os resultados das consultas por similaridade. Conseqüentemente, desdobra novos caminhos para realçar a concepção de sistemas CBIR / The retrieval of images by visual content relies on a feature extractor to provide the most meaningful intrinsic characteristics (features) from the data, and a distance function to quantify the similarity between them. A challenge in this field supporting content-based image retrieval (CBIR) to answer similarity queries is how to best integrate these two key aspects. There are plenty of researching on algorithms for feature extraction of images. However, little attention have been paid to the importance of the use of a well-suited distance function associated to a feature extractor. This Master Dissertation was conceived to fill in this gap. Therefore, herein it was investigated the behavior of different distance functions regarding distinct feature vector types. The three main types of image features were evaluated, regarding color distribution, texture and shape. It was also proposed two new techniques to perform feature selection over the feature vectors, in order to improve the precision when answering similarity queries. The first technique employed statistical association rules and achieve up to 38% gain in precision, while the second one employing the Shannon entropy achieved 71%, while siginificantly reducing the size of the feature vector. This work also showed that the proper use of a distance function effectively improves the similarity query results. Therefore, it opens new ways to enhance the acceptance of CBIR systems

La recherche d'images basée sur le contenu visuel est un domaine très actif de la vision par ordinateur, car le nombre de bases d'images disponibles ne cesse d'augmenter.L’objectif de ce type d’approche est de retourner les images les plus proches d'une requête donnée en terme de contenu visuel.Notre travail s'inscrit dans un contexte applicatif spécifique qui consiste à indexer des petites bases d'images expertes sur lesquelles nous n'avons aucune connaissance a priori.L’une de nos contributions pour palier ce problème consiste à choisir un ensemble de descripteurs visuels et de les placer en compétition directe. Nous utilisons deux stratégies pour combiner ces caractéristiques : la première, est pyschovisuelle, et la seconde, est statistique.Dans ce contexte, nous proposons une approche adaptative non supervisée, basée sur les sacs de mots et phrases visuels, dont le principe est de sélectionner les caractéristiques pertinentes pour chaque point d'intérêt dans le but de renforcer la représentation de l'image.Les tests effectués montrent l'intérêt d'utiliser ce type de méthodes malgré la domination des méthodes basées réseaux de neurones convolutifs dans la littérature.Nous proposons également une étude, ainsi que les résultats de nos premiers tests concernant le renforcement de la recherche en utilisant des méthodes semi-interactives basées sur l’expertise de l'utilisateur. / Image Retrieval is still a very active field of image processing as the number of available image datasets continuously increases.One of the principal objectives of Content-Based Image Retrieval (CBIR) is to return the most similar images to a given query with respect to their visual content.Our work fits in a very specific application context: indexing small expert image datasets, with no prior knowledge on the images. Because of the image complexity, one of our contributions is the choice of effective descriptors from literature placed in direct competition.Two strategies are used to combine features: a psycho-visual one and a statistical one.In this context, we propose an unsupervised and adaptive framework based on the well-known bags of visual words and phrases models that select relevant visual descriptors for each keypoint to construct a more discriminative image representation.Experiments show the interest of using this this type of methodologies during a time when convolutional neural networks are ubiquitous.We also propose a study about semi interactive retrieval to improve the accuracy of CBIR systems by using the knowledge of the expert users.

Image segmentation is one of the most difficult tasks in image processing. Segmentation algorithms are generally based on searching a region where pixels share similar gray level intensity and satisfy a set of defined criteria. However, the segmented region cannot be used directly for partial image retrieval. In this dissertation, a Contour Based Image Structure (CBIS) model is introduced. In this model, images are divided into several objects defined by their bounding contours. The bounding contour structure allows individual object extraction, and partial object matching and retrieval from a standard CBIS image structure. The CBIS model allows the representation of 3D objects by their bounding contours which is suitable for parallel implementation particularly when extracting contour features and matching them for 3D images require heavy computations. This computational burden becomes worse for images with high resolution and large contour density. In this essence we designed two parallel algorithms; Contour Parallelization Algorithm (CPA) and Partial Retrieval Parallelization Algorithm (PRPA). Both algorithms have considerably improved the performance of CBIS for both contour shape matching as well as partial image retrieval. To improve the effectiveness of CBIS in segmenting images with inhomogeneous backgrounds we used the phase congruency invariant features of Fourier transform components to highlight boundaries of objects prior to extracting their contours. The contour matching process has also been improved by constructing a fuzzy contour matching system that allows unbiased matching decisions. Further improvements have been achieved through the use of a contour tailored Fourier descriptor to make translation and rotation invariance. It is proved to be suitable for general contour shape matching where translation, rotation, and scaling invariance are required. For those images which are hard to be classified by object contours such as bacterial images, we define a multi-level cosine transform to extract their texture features for image classification. The low frequency Discrete Cosine Transform coefficients and Zenike moments derived from images are trained by Support Vector Machine (SVM) to generate multiple classifiers.

La recherche d’images par contenu (CBIR) est une discipline de l’informatique qui vise à structurer automatiquement les collections d’images selon des critères visuels. Les fonctionnalités proposées couvrent notamment l’accès efficace aux images dans une grande base de données d’images ou l’identification de leur contenu par des outils de détection et de reconnaissance d’objets. Ils ont un impact sur une large gamme de domaines qui manipulent ce genre de données, telles que le multimedia, la culture, la sécurité, la santé, la recherche scientifique, etc.Indexer une image à partir de son contenu visuel nécessite d’abord de produire un résumé visuel de ce contenu pour un usage donné, qui sera l’index de cette image dans la collection. En matière de descripteurs d’images, la littérature est désormais trés riche: plusieurs familles de descripteurs existent, et dans chaque famille de nombreuses approches cohabitent. Bon nombre de descripteurs ne décrivant pas la même information et n’ayant pas les mêmes propriétés d’invariance, il peut être pertinent de les combiner de manière à mieux décrire le contenu de l’image. Cette combinaison peut être mise en oeuvre de différentes manières, selon les descripteurs considérés et le but recherché. Dans cette thése, nous nous concentrons sur la famille des descripteurs locaux, avec pour application la recherche d’images ou d’objets par l’exemple dans une collection d’images. Leurs bonnes propriétés les rendent très populaires pour la recherche, la reconnaissance et la catégorisation d'objets et de scènes. Deux directions de recherche sont étudiées:Combinaison de caractéristiques pour la recherche d’images par l’exemple: Le coeur de la thèse repose sur la proposition d’un modèle pour combiner des descripteurs de bas niveau et génériques afin d’obtenir un descripteur plus riche et adapté à un cas d’utilisation donné tout en conservant la généricité afin d’indexer différents types de contenus visuels. L’application considérée étant la recherche par l’exemple, une autre difficulté majeure est la complexité de la proposition, qui doit correspondre à des temps de récupération réduits, même avec de grands ensembles de données. Pour atteindre ces objectifs, nous proposons une approche basée sur la fusion d'index inversés, ce qui permet de mieux représenter le contenu tout en étant associé à une méthode d’accès efficace.Complémentarité des descripteurs: Nous nous concentrons sur l’évaluation de la complémentarité des descripteurs locaux existant en proposant des critères statistiques d’analyse de leur répartition spatiale dans l'image. Ce travail permet de mettre en évidence une synergie entre certaines de ces techniques lorsqu’elles sont jugées suffisamment complémentaires. Les critères spatiaux sont exploités dans un modèle de prédiction à base de régression linéaire, qui a l'avantage de permettre la sélection de combinaisons de descripteurs optimale pour la base considérée mais surtout pour chaque image de cette base. L'approche est évaluée avec le moteur de recherche multi-index, où il montre sa pertinence et met aussi en lumière le fait que la combinaison optimale de descripteurs peut varier d'une image à l'autre.En outre, nous exploitons les deux propositions précédentes pour traiter le problème de la recherche d'images inter-domaines, correspondant notamment à des vues multi-source et multi-date. Deux applications sont explorées dans cette thèse. La recherche d’images inter-domaines est appliquée aux collections photographiques culturelles numérisées d’un musée, où elle démontre son efficacité pour l’exploration et la valorisation de ces contenus à différents niveaux, depuis leur archivage jusqu’à leur exposition ou ex situ. Ensuite, nous explorons l’application de la localisation basée image entre domaines, où la pose d’une image est estimée à partir d’images géoréférencées, en retrouvant des images géolocalisées visuellement similaires à la requête / Content-Based Image Retrieval (CBIR) is a discipline of Computer Science which aims at automatically structuring image collections according to some visual criteria. The offered functionalities include the efficient access to images in a large database of images, or the identification of their content through object detection and recognition tools. They impact a large range of fields which manipulate this kind of data, such as multimedia, culture, security, health, scientific research, etc.To index an image from its visual content first requires producing a visual summary of this content for a given use, which will be the index of this image in the database. From now on, the literature on image descriptors is very rich; several families of descriptors exist and in each family, a lot of approaches live together. Many descriptors do not describe the same information and do not have the same properties. Therefore it is relevant to combine some of them to better describe the image content. The combination can be implemented differently according to the involved descriptors and to the application. In this thesis, we focus on the family of local descriptors, with application to image and object retrieval by example in a collection of images. Their nice properties make them very popular for retrieval, recognition and categorization of objects and scenes. Two directions of research are investigated:Feature combination applied to query-by-example image retrieval: the core of the thesis rests on the proposal of a model for combining low-level and generic descriptors in order to obtain a descriptor richer and adapted to a given use case while maintaining genericity in order to be able to index different types of visual contents. The considered application being query-by-example, another major difficulty is the complexity of the proposal, which has to meet with reduced retrieval times, even with large datasets. To meet these goals, we propose an approach based on the fusion of inverted indices, which allows to represent the content better while being associated with an efficient access method.Complementarity of the descriptors: We focus on the evaluation of the complementarity of existing local descriptors by proposing statistical criteria of analysis of their spatial distribution. This work allows highlighting a synergy between some of these techniques when judged sufficiently complementary. The spatial criteria are employed within a regression-based prediction model which has the advantage of selecting the suitable feature combinations globally for a dataset but most importantly for each image. The approach is evaluated within the fusion of inverted indices search engine, where it shows its relevance and also highlights that the optimal combination of features may vary from an image to another.Additionally, we exploit the previous two proposals to address the problem of cross-domain image retrieval, where the images are matched across different domains, including multi-source and multi-date contents. Two applications of cross-domain matching are explored. First, cross-domain image retrieval is applied to the digitized cultural photographic collections of a museum, where it demonstrates its effectiveness for the exploration and promotion of these contents at different levels from their archiving up to their exhibition in or ex-situ. Second, we explore the application of cross-domain image localization, where the pose of a landmark is estimated by retrieving visually similar geo-referenced images to the query images

L'objectif de ce travail est de proposer une solution de reconnaissance de l'impact émotionnel des images en se basant sur les techniques utilisées en recherche d'images par le contenu. Nous partons des résultats intéressants de cette architecture pour la tester sur une tâche plus complexe. La tâche consiste à classifier les images en fonction de leurs émotions que nous avons définies "Négative", "Neutre" et "Positive". Les émotions sont liées aussi bien au contenu des images, qu'à notre vécu. On ne pourrait donc pas proposer un système de reconnaissance des émotions performant universel. Nous ne sommes pas sensible aux mêmes choses toute notre vie : certaines différences apparaissent avec l'âge et aussi en fonction du genre. Nous essaierons de nous affranchir de ces inconstances en ayant une évaluation des bases d'images la plus hétérogène possible. Notre première contribution va dans ce sens : nous proposons une base de 350 images très largement évaluée. Durant nos travaux, nous avons étudié l'apport de la saillance visuelle aussi bien pendant les expérimentations subjectives que pendant la classification des images. Les descripteurs, que nous avons choisis, ont été évalués dans leur majorité sur une base consacrée à la recherche d'images par le contenu afin de ne sélectionner que les plus pertinents. Notre approche qui tire les avantages d'une architecture bien codifiée, conduit à des résultats très intéressants aussi bien sur la base que nous avons construite que sur la base IAPS, qui sert de référence dans l'analyse de l'impact émotionnel des images. / The goal of this work is to propose an efficient approach for emotional impact recognition based on CBIR techniques (descriptors, image representation). The main idea relies in classifying images according to their emotion which can be "Negative", "Neutral" or "Positive". Emotion is related to the image content and also to the personnal feelings. To achieve our goal we firstly need a correct assessed image database. Our first contribution is about this aspect. We proposed a set of 350 diversifed images rated by people around the world. Added to our choice to use CBIR methods, we studied the impact of visual saliency for the subjective evaluations and interest region segmentation for classification. The results are really interesting and prove that the CBIR methods are usefull for emotion recognition. The chosen desciptors are complementary and their performance are consistent on the database we have built and on IAPS, reference database for the analysis of the image emotional impact.

Submitted by Aelson Maciera (aelsoncm@terra.com.br) on 2017-09-13T18:11:26Z No. of bitstreams: 1 DissACC.pdf: 1668214 bytes, checksum: 82bf6ff6918613ce74cc691a951a22b0 (MD5) / Approved for entry into archive by Ronildo Prado (bco.producao.intelectual@gmail.com) on 2018-01-25T17:53:52Z (GMT) No. of bitstreams: 1 DissACC.pdf: 1668214 bytes, checksum: 82bf6ff6918613ce74cc691a951a22b0 (MD5) / Approved for entry into archive by Ronildo Prado (bco.producao.intelectual@gmail.com) on 2018-01-25T17:54:04Z (GMT) No. of bitstreams: 1 DissACC.pdf: 1668214 bytes, checksum: 82bf6ff6918613ce74cc691a951a22b0 (MD5) / Made available in DSpace on 2018-01-25T18:00:35Z (GMT). No. of bitstreams: 1 DissACC.pdf: 1668214 bytes, checksum: 82bf6ff6918613ce74cc691a951a22b0 (MD5) Previous issue date: 2017-04-18 / Não recebi financiamento / One of the ways to recover images in a database is through similarity queries. Using characteristics extracted from these images, such as color, shape or texture, this work seeks to identify similarities to a central query element. However, the results may be very similar to each other, which is not always the expected result. In addition to the redundancy in the results, the problem of the ’semantic gap’, which is a divergence in the evaluation of similarity between images performed by the computer considering its numerical representation (low level characteristics) and the human perception about the image (high level characteristics). In order to improve the quality of the results, we sought to minimize the issue of redundancy and the ’semantic gap’ through the use of more than one descriptor in queries for similarity. We sought to explore the inclusion of diversity using one descriptor to treat similarity and another descriptor to treat diversity, more generally a metric space for similarity and another for diversity. For the implementation of the query by similarity was used the consultation to several neighbors closer. Considering that the descriptors may be distinct and one of them may have greater numerical representativeness, it was necessary to do the normalization, considering the methods of normalization by the greater distance and normalization by the greater approximate distance with balancing by the intrinsic dimension. An exhaustive search algorithm was used to perform the tests. The experiments were carried out in a classified database. To evaluate the semantic quality of the results, a measure was proposed that evaluates the inclusion of diversity considering the diversity present in the query only considering the similarity and the maximum diversity that can be included. A comparison was made between the result obtained and the considered ideal, which refers to the value of l defined by the user himself. By comparing the results obtained with the results obtained in the queries for a single descriptor, the evaluation of the included diversity followed the trend of l, which allows to say that normalization and balancing is necessary. In addition, it is intended in the future to study new ways of normalizing. / Uma das formas para se recuperar imagens em banco de dados, é através de consultas por similaridade. Utilizando características extraídas dessas imagens, como cor, forma ou textura, busca-se identificar semelhanças a um elemento central de consulta. No entanto, os resultados nas consultas podem ser muito semelhantes entre si, o que nem sempre é o resultado esperado. Além da redundância nos resultados, deve-se destacar o problema do ‘gap semântico’, que é a divergência na avaliação da similaridade entre imagens realizada pelo computador considerando a sua representação numérica (características de baixo nível) e a percepção humana sobre a imagem (características de alto nível). Com o objetivo de melhorar a qualidade dos resultados nas consultas buscou-se minimizar a questão da redundância e do ‘gap semântico’ através da utilização de mais de um descritor nas consultas por similaridade. Buscou-se explorar a inclusão de diversidade utilizando-se um descritor para tratar a similaridade e outro descritor para tratar a diversidade, mais genericamente, um espaço métrico para similaridade e outro para a diversidade. Para a implementação da consulta por similaridade utilizou-se a consulta aos vizinhos diversos mais próximos. Considerando-se que os descritores utilizados podem ser distintos e que um deles possa ter maior representatividade numérica do que o outro, foi necessário fazer a normalização, sendo considerados os métodos da normalização pela maior distância e normalização pela maior distancia aproximada com balanceamento pela dimensão intrínseca. Para a realização dos testes utilizou-se um algoritmo de busca exaustiva. Os experimentos foram realizados em uma base de dados classificada. Para avaliar a qualidade semântica dos resultados foi proposta uma medida que avalia a inclusão de diversidade considerando a diversidade presente na consulta apenas considerando a similaridade e a diversidade máxima que pode ser incluída. Foi feita uma comparação entre o resultado obtido e o considerado ideal, que refere-se ao valor de l definido pelo próprio usuário. Comparando-se os resultados alcançados com os resultados obtidos nas consultas para um único descritor, a avaliação da diversidade incluída acompanhou a tendência de l, o que permite dizer que a normalização e balanceamento é necessário. Além disso, pretende-se futuramente estudar novas formas de normalizar.

Made available in DSpace on 2016-06-02T19:06:11Z (GMT). No. of bitstreams: 1 5726.pdf: 4603491 bytes, checksum: 0fe3fa824a018f481106303c4816bf07 (MD5) Previous issue date: 2013-08-23 / Financiadora de Estudos e Projetos / Content-Based Image Retrieval systems use visual information like color, shape and texture to represent images in feature vectors. The numerical representation found for the images is used in query execution through a metric to evaluate the distance between vectors. In general, there is an inconsistency in the evaluation of similarity between images according to human perception and the results computed by CBIR systems, which is called Semantic Gap. One way to overcome this problem is by the addition of a diversity factor in query execution, allowing the user to specify a degree of dissimilarity between the resulting images and changing the query result. Adding diversity in consultation, however, requires high computational cost and the reduction of possible subsets to be analyzed is a difficult task to be understood by the user. This masters degree thesis aims to make use of Visual Data Mining techniques applied to queries in CBIR systems, improving the interpretability of the measure of similarity and diversity, as well as the relevance of the result according to the judgment and prior knowledge of the user. The user takes an active role in the retrieval of images by their content, guiding its result and, consequently, reducing the Semantic Gap. Additionally, a better understanding of the diversity and similarity factors involved in the query is supported by visualization and interaction techniques. / Sistemas de recuperação de imagens por conteúdo (do Inglês, Content-Based Image Retrieval - CBIR) utilizam informações visuais de cor, forma e textura para representar as imagens em vetores de características. A representação numérica encontrada para as imagens é utilizada na execução da consulta através de uma métrica que avalie a distância entre os vetores. Em geral, existe uma inconsistência entre a percepção do ser humano na avaliação de similaridade entre imagens se comparada com a computada por sistemas CBIR, sendo esta descontinuidade denominada Gap Semântico. Adicionar um fator de diversidade na consulta tem-se mostrado como uma maneira de superar este problema, permitindo que o usuário especifique o grau de dissimilaridade entre as imagens resultantes e altere o resultado da consulta. Adicionar diversidade em consulta, no entanto, requer alto custo computacional e a redução das possibilidades de conjuntos para resposta é de difícil entendimento para o usuário. Este trabalho de mestrado propôs a utilização de técnicas de Mineração Visual de Dados (MVD) aplicadas sobre consultas em sistemas CBIR, melhorando a interpretabilidade da medida de similaridade e diversidade, assim como a relevância do resultado obtido. O usuário passa a exercer um papel ativo na consulta por conteúdo de imagens, permitindo que o mesmo dirija o processo, aproximando o resultado ao esperado pela cognição humana e reduzindo o gap semântico.

Made available in DSpace on 2015-05-14T12:36:54Z (GMT). No. of bitstreams: 1 arquivototal.pdf: 2499209 bytes, checksum: 80d399f8f00f3e82d2a3b34e52fd6b05 (MD5) Previous issue date: 2009-08-27 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - CAPES / As the processing power of computers increases, the quantity and complexity of stored data have growing in the same way, requiring more sophisticated mechanisms to accomplish retrieval with efficacy and efficiency over these information. In image processing, it has become common the retrieval based on its own content, namely Content-Based Image Retrieval (CBIR), which eliminates the need to place additional annotations as textual descriptions and keywords registered by an observer. The purpose of this work is the development of an image retrieval mechanism based on shape recognition. The mechanism consists in (1) compute the Full Curvature Scale Space (FullCSS) image descriptors; and (2) apply over them a lossless compression method objecting to (3) classify these descriptors and retrieve the corresponding images. The FullCSS descriptors register the curvature variations on the image contour indicating the degree and the signal of these variations, which allow identifying where the curvature is concave or convex. The adopted compression method uses the Prediction by Partial Matching (PPM) compression model, which has been successfully used in other works to classify texture images. The results obtained show that this novel approach is able to reach competitive levels of efficacy and efficiency when compared to other works recently developed in this same area. / Com o aumento do poder de processamento dos computadores, cresceu também a quantidade e complexidade dos dados armazenados, exigindo mecanismos cada vez mais sofisticados para se conseguir uma recuperação eficaz e eficiente destas informações. No caso do processamento de imagens, tem se tornado comum a recuperação baseada em seu próprio conteúdo, ou seja, Recuperação de Imagem Baseada em Conteúdo (Content-Based Image Retrieval CBIR), eliminando a necessidade de anotações adicionais como descrições textuais e palavras-chave registradas por um observador. A proposta deste trabalho é o desenvolvimento de um mecanismo de recuperação de imagens através do reconhecimento de sua forma. O mecanismo consiste em (1) calcular os descritores Full Curvature Scale Space (FullCSS) das imagens; e (2) aplicar sobre eles um método de compressão sem perdas com a finalidade de (3) classificar esses descritores e recuperar as imagens correspondentes. Os descritores FullCSS registram as variações na curvatura do contorno da imagem indicando o grau e o sinal dessas variações, permitindo identificar onde a curvatura é côncava ou convexa. O método de compressão adotado utiliza o modelo de compressão Prediction by Partial Matching (PPM), utilizado com sucesso em outros trabalhos para classificar imagens de texturas. Os resultados obtidos indicam que esta abordagem inovadora é capaz de atingir níveis competitivos de eficácia e eficiência quando comparada a outros trabalhos atualmente desenvolvidos nesta mesma área.

La mutualisation et l'externalisation de données concernent de nombreux domaines y compris celui de la santé. Au-delà de la réduction des coûts de maintenance, l'intérêt est d'améliorer la prise en charge des patients par le déploiement d'outils d'aide au diagnostic fondés sur la réutilisation des données. Dans un tel environnement, la sécurité des données (confidentialité, intégrité et traçabilité) est un enjeu majeur. C'est dans ce contexte que s'inscrivent ces travaux de thèse. Ils concernent en particulier la sécurisation des techniques de recherche d'images par le contenu (CBIR) et de « machine learning » qui sont au c'ur des systèmes d'aide au diagnostic. Ces techniques permettent de trouver des images semblables à une image requête non encore interprétée. L'objectif est de définir des approches capables d'exploiter des données externalisées et sécurisées, et de permettre à un « cloud » de fournir une aide au diagnostic. Plusieurs mécanismes permettent le traitement de données chiffrées, mais la plupart sont dépendants d'interactions entre différentes entités (l'utilisateur, le cloud voire un tiers de confiance) et doivent être combinés judicieusement de manière à ne pas laisser fuir d'information lors d'un traitement.Au cours de ces trois années de thèse, nous nous sommes dans un premier temps intéressés à la sécurisation à l'aide du chiffrement homomorphe, d'un système de CBIR externalisé sous la contrainte d'aucune interaction entre le fournisseur de service et l'utilisateur. Dans un second temps, nous avons développé une approche de « Machine Learning » sécurisée fondée sur le perceptron multicouches, dont la phase d'apprentissage peut être externalisée de manière sûre, l'enjeu étant d'assurer la convergence de cette dernière. L'ensemble des données et des paramètres du modèle sont chiffrés. Du fait que ces systèmes d'aides doivent exploiter des informations issues de plusieurs sources, chacune externalisant ses données chiffrées sous sa propre clef, nous nous sommes intéressés au problème du partage de données chiffrées. Un problème traité par les schémas de « Proxy Re-Encryption » (PRE). Dans ce contexte, nous avons proposé le premier schéma PRE qui permet à la fois le partage et le traitement des données chiffrées. Nous avons également travaillé sur un schéma de tatouage de données chiffrées pour tracer et vérifier l'intégrité des données dans cet environnement partagé. Le message tatoué dans le chiffré est accessible que l'image soit ou non chiffrée et offre plusieurs services de sécurité fondés sur le tatouage. / Cloud computing has emerged as a successful paradigm allowing individuals and companies to store and process large amounts of data without a need to purchase and maintain their own networks and computer systems. In healthcare for example, different initiatives aim at sharing medical images and Personal Health Records (PHR) in between health professionals or hospitals with the help of the cloud. In such an environment, data security (confidentiality, integrity and traceability) is a major issue. In this context that these thesis works, it concerns in particular the securing of Content Based Image Retrieval (CBIR) techniques and machine learning (ML) which are at the heart of diagnostic decision support systems. These techniques make it possible to find similar images to an image not yet interpreted. The goal is to define approaches that can exploit secure externalized data and enable a cloud to provide a diagnostic support. Several mechanisms allow the processing of encrypted data, but most are dependent on interactions between different entities (the user, the cloud or a trusted third party) and must be combined judiciously so as to not leak information. During these three years of thesis, we initially focused on securing an outsourced CBIR system under the constraint of no interaction between the users and the service provider (cloud). In a second step, we have developed a secure machine learning approach based on multilayer perceptron (MLP), whose learning phase can be outsourced in a secure way, the challenge being to ensure the convergence of the MLP. All the data and parameters of the model are encrypted using homomorphic encryption. Because these systems need to use information from multiple sources, each of which outsources its encrypted data under its own key, we are interested in the problem of sharing encrypted data. A problem known by the "Proxy Re-Encryption" (PRE) schemes. In this context, we have proposed the first PRE scheme that allows both the sharing and the processing of encrypted data. We also worked on watermarking scheme over encrypted data in order to trace and verify the integrity of data in this shared environment. The embedded message is accessible whether or not the image is encrypted and provides several services.

Este trabalho foi desenvolvido com o intuito de se estabelecer uma metodologia para a classificação das cores de imagens digitais em cores perceptuais para se gerar um vetor de características que permita recuperar imagens através de seu conteúdo em uma base de dados. Em trabalhos e estudos correlatos analisados, as metodologias de agrupamento das diversas cores possíveis de uma imagem não permitem uma associação entre a cor digitalizada e a cor percebida por seres humanos. Estudos mostram que a maioria das culturas humanas associam às cores apenas onze termos: vermelho, amarelo, violeta, azul, verde, rosa, marrom, preto, branco, laranja e cinza. Este trabalho propõe, portanto, uma metodologia baseada em regras da lógica fuzzy, que permite associar a todas as possíveis cores de imagens digitais uma das onze cores culturais definidas, criando assim um histograma perceptual de cores. Isso permitiu a geração de um vetor de características para a recuperação de imagens baseada em conteúdo em uma base de dados. / This work aims at establishing a digital image classification methodology based on perceptual colors, by generating a feature vector that allows retrieving images from a database by their content. In related works the methodologies of grouping the diverse possible colors of an image do not allow associate digitized colors and those colors perceived by human beings. Studies show that the majority of human being culture associates only eleven terms to all the possible colors: red, yellow, blue, green, pink, brown, black, white, purple, orange and gray. This work purpose a methodology based on fuzzy logic that allows to associate the eleven cultural color terms with all of digitized colors by a perceptual color histogram. The image color quantization generates a feature vector used for content-based image retrieval. The results show that it is possible to use the perceptual color histogram for CBIR and in the semantic gap reduction.