Ce travail porte sur une étude comparative des huiles végétales (esters naturels) issues de différentes graines, esters synthétiques, et minérales sur la base des caractéristiques de génération et de propagation des streamers qui s’y développent, des tensions de claquage ainsi que des décharges se propageant sur un isolant solide en présence de ces huiles. Il ressort des résultats obtenus qu’en géométrie pointe – plan, sous tension continu, la tension de génération en polarité positive est toujours supérieure à celle mesurée en polarité négative et elle est plus élevée dans les huiles esters que dans les huiles minérales quelle que soit la polarité de l’électrode pointe. En ce qui concerne la propagation de streamers sous tension impulsionnelle de foudre, les longueurs finales (ou longueurs d’arrêt) Lf des streamers positifs sont environ10 fois supérieures à celles des streamers négatifs. Pour atteindre la même Lf, il faut beaucoup plus de tension en polarité négative qu’en polarité positive. Par conséquent, les tenues diélectriques sont plus élevées en polarité négative qu’en polarité positive. D’autre part, Lf est plus élevée dans les huiles végétales que dans certaines huiles minérales. La tension de claquage des huiles isolantes, déterminée selon la norme CEI 60156, suit généralement une distribution Normale. Les probabilités de claquage peuvent alors être déterminées suivant la distribution Normale en prenant un échantillonnage suffisamment grand (n>30). Il est également montré que les huiles végétales et l’ester synthétique présentent des tensions de claquage (U1%, U10% et U50%) plus élevées que les huiles minérales sous tension alternative. La tension de claquage moyenne est aussi plus élevée dans les esters que dans les huiles minérales sous tension impulsionnelle de foudre. La morphologie, la longueur d’arrêt et le courant des décharges se propageant sur un isolant solide (carton) immergé dans les liquides étudiés, dépendent de l’amplitude et de la polarité de la tension, de l’épaisseur du solide et de la nature des huiles. Lf augmente quasi-linéairement avec la tension appliquée et diminue lorsque l’épaisseur du solide augmente. Pour une tension et une épaisseur données, les décharges qui se développent sur l’interface carton/huile végétale sont plus longues que celles sur l’interface carton/huile minérale. Par ailleurs, Lf est plus élevé en polarité positive qu’en polarité négative, ce qui indique que la polarité la plus contraignante est la polarité positive. Ila été également mis en évidence l’existence d’une décharge secondaire de signe opposé à celui de la tension appliquée, résultat de l’accumulation de charges d’espace à la surface de l’isolant solide. Enfin, la dimension fractale D des décharges diminue lorsque l’épaisseur du solide augmente. Plus l’épaisseur du solide est grande, plus le degré de ramification des décharges est moins important. Pour une tension et une épaisseur données, D est généralement plus élevée dans l’huile minérale que dans l’huile végétale. / This work is aimed at the comparative study of vegetable oils (natural esters) issued from different seeds, synthetic ester oil and mineral oils, based on the characteristics of generation and propagation of streamers, breakdown voltages and creeping discharges propagating over solid insulator (pressboard) immersed in these oils. It has been shown that, in a point – plane electrode geometry, the DC threshold voltage Us is higher when the point is positive than when it’s negative; Us is higher in vegetable oils than in mineral oils for both polarities of voltage. Under lightning impulse voltage, the streamers are filamentary in all different tested oils whatever the polarity of voltage. The final (or stopping) length Lf of streamers are about ten times higher when the point is positive than when it’s negative. The voltage required to reach the same Lf is higher with a negative point than with a positive point.Thus the positive polarity is unfavourable for the industrial applications. In addition, Lf is longer in vegetable oils than in some mineral oils. The breakdown voltage of insulating oils, measured according to IEC 60156 standard specifications generally follows a normal distribution. The probability of breakdown can hence be determined according to the normal distribution with a large amount of samples (n>30). So, vegetable oils and synthetic ester have breakdown voltages (U1%, U10% and U50%) higher than mineral oils under AC; and the mean breakdown voltage is also higher in esters oils than in minerals oils under lightning impulse voltage. The morphology, final length Lf and current of discharges propagating over a solid insulator (pressboard) immersed in the liquid tested, depend on the amplitude and polarity of the voltage, the thickness of insulator as well as the nature of insulating oils. It has been shown that Lf increases quasi-linearly with the applied voltage and decreases when the thickness of insulator increases. For a given applied voltage and thickness of insulator, Lf is higher with a pressboard/vegetable oil interface than with a pressboard/mineral oil interface.In addition, Lf is also longer when the point is positive than when it’s negative. The recorded currents revealed the existence of a secondary discharge of opposite sign to that of the applied voltage. This phenomenon is due to the accumulation of charges at the surface of insulator.Finally, the fractal dimension D of creeping discharges decreases when the thickness of insulator increases. For a given applied voltage and thickness of solid, D is generally higher in mineral oils than in vegetable oils.

Le présent travail porte sur une étude comparative des caractéristiques optiques et électriques des décharges glissantes se propageant aux interfaces solide/gaz sur des isolateurs de verre, de Bakélite et de résine époxy en présence des gaz N2, CO2 et SF6 et des mélanges SF6/N2 et SF6/CO2, sous tension continue et alternative (50 Hz), en géométrie pointe - plan. L’objectif est de mieux comprendre les mécanismes impliqués dans l’initiation des décharges partielles et leur évolution et développement en décharges surfaciques (glissantes) lesquelles peuvent conduire au contournement des composants et systèmes haute tension. Les résultats obtenus montrent que la morphologie et la longueur finale (d’arrêt ou d’extension maximale) des décharges surfaciques dépendent de la forme et de l’amplitude de la tension, de l’épaisseur et de la nature du solide isolant, du type du gaz/mélange et de sa pression. Il est montré que la longueur finale des décharges Lf augmente quasi-linéairement avec la tension. Lf diminue lorsque la pression du gaz et/ou l’épaisseur du solide augmentent. Cette longueur est plus courte dans le SF6 que dans le CO2 ou le N2 ; et elle diminue significativement lorsque le taux du SF6 dans le mélange de gaz augmente. Par ailleurs, pour une tension donnée, Lf augmente avec la constante diélectrique de l’isolant solide. La longueur finale des décharges est nettement plus élevée sous tension alternative que sous tension continue. La morphologie des décharges glissantes générées sous tension continue et alternative est généralement non radiale; leur orientation est fortement influencée par la présence des charges d’espace présentes ou déposées sur la surface de l’isolateur. Une analyse fractale des décharges glissantes obtenues expérimentalement sous tension continue est également réalisée et une corrélation entre la dimension fractale, la pression du gaz, la constante diélectrique et l’épaisseur du matériau solide est mise en évidence. En particulier, la dimension fractale augmente lorsque la constante diélectrique augmente et/ou l’épaisseur de l’isolant solide diminue et/ou la pression du gaz diminue. / This work deals with a comparative study of optical and electrical characteristics of creeping discharges propagating at solid/gas interfaces on insulators made of glass, Bakelite and epoxy resin in the presence of N2, CO2 and SF6 gases and SF6/N2 SF6/CO2 mixtures, under DC and AC (50 Hz) voltage, using a point - plane electrode arrangement. The objective is to better understand the mechanisms involved in the initiation of partial discharges and their evolution and development into discharges surface (creeping discharges) that can lead to flashover of components and high voltage systems. The results show that the morphology and final length (maximum extension or stopping length) depend on the shape and amplitude of the voltage, the thickness and the nature of the solid insulator, type of gas / mixture and its pressure. It is shown that the final length Lf increases quasi-linearly with the voltage. Lf decreases as the gas pressure increases and/ or the thickness of the solid increases. Lf is shorter in SF6 than in CO2 or N2, and it decreases significantly when the rate of SF6 in the gas mixture increases. Moreover, for a given voltage, Lf increases with the dielectric constant of the solid insulation. The final length of the discharge is much higher under AC voltage than under DC voltage. The morphology of creeping discharges generated under DC and AC voltage is generally not radial; and their orientation is strongly influenced by the presence of space charges present or deposited on the surface of the insulator. A fractal analysis of creeping discharges experimentally obtained under DC voltage is also carried out and a correlation between the fractal dimension, the pressure of the gas, the dielectric constant and the thickness of the solid material is highlighted. In particular, the fractal dimension increases when the dielectric constant increases and / or thickness of the solid insulator decreases and / or pressure of the gas decreases.

The key objective of this thesis is the outline of an alternative capital market modeling framework, the Volatility Asset Pricing Model, VAPM, inspired by the innovative dual approach of Mandelbrot and Hudson using the method based on synthesis of two seemingly antagonistic factors -- the volatility of market prices and their serial dependence determining the capital markets' dynamics. The pilot tests of this model in various periods using the market index as well as a portfolio of selected securities delivered generally satisfactory results. Firstly, the work delivers a brief recapitulation regarding the concepts of a consumer/investor choice under general conditions of hypothetical certainty. Secondly, this outline is then followed by a description of the "classical" methodologies in the risky environment of uncertainty, with assessment of their corresponding key models, i.e. the CAPM, SIM, MIM, APTM, etc., notwithstanding results of the related testing approaches. Thirdly, this assessment is based on evaluation of the underlying doctrine of Efficient Market Hypothesis in relation to the so called Random Walk Model. Fourthly, in this context the work also offers a brief exposure to a few selected tests of these contraversial concepts. Fifthly, the main points of conteporary approaches such as the Fractal Dimension and the Hurst Exponent in the dynamic framework of information entropy are subsequently described as the theoretical tools leading to development of the abovementioned model VAPM. The major contribution of this thesis is considered its attempt to apply the abovementioned concepts in practice, with the intention to possibly inspire a further analytical research.

Due to the increasing amount and complexity of data processed in real world applications, similarity search became a vital task to store and retrieve such data. However, missing attribute values are very frequent and metric access methods (MAMs), designed to support similarity search, do not operate on datasets when attribute values are missing. Currently, the approach to use the existing indexing techniques on datasets with missing attribute values just use an indicator to identify the missing values and employ a traditional indexing technique. Although, this approach can be applied over multidimensional indexing techniques, it is impractical for metric access methods. This dissertation presents the results of a research conducted to identify and deal with the issues related to indexing and querying datasets with missing values in metric spaces. An empirical analysis of the metric access methods when applied on incomplete datasets leads us to identify two main issues: distortion of the internal structure of the index when data are missing at random and skew of the index structure when data are not missing at random. Based on those findings, a new variant of the Slim-tree access method, called Hollow-tree, is presented. It employs new techniques that are capable to handle missing data issues when missingness is ignorable. The first technique includes a set of indexing policies that allow to index objects with missing attribute values and prevent distortions to occur in the internal structure of the indexes. The second technique targets the similarity queries to improve the query performance over incomplete datasets. This technique employs the fractal dimension of the dataset and the local density around the query object to estimate an ideal radius able to achieve an accurate query answer, considering data with missing values as a potential response. Results from experiments with a variety of real and synthetic datasets show that Hollow-tree achieves nearly 100% of precision and recall for Range queries and more than 90% for k Nearest Neighbor queries, while Slim-tree access method deteriorates with the increasing amount of missing values. The results confirm that the indexing technique helps to establish consistency in the index structure and the searching technique achieves a remarkable performance. When combined, the new techniques allow to explore properly all the available data even with high amounts of missing attribute values. As they are independent of the underlying access method, they can be adopted by a broad range of metric access methods, allowing to extend the class of MAMs. / O crescimento em quantidade e complexidade dos dados processados e armazenados torna a busca por similaridade uma tarefa fundamental para tratar esses dados. No entanto, atributos faltantes ocorrem freqüentemente, inviabilizando os métodos de acesso métricos (MAMs) projetados para apoiar a busca por similaridade. Assim, técnicas de tratamento de dados faltantes precisam ser desenvolvidas. A abordagem mais comum para executar as técnicas de indexação existentes sobre conjuntos de dados com valores faltantes é usar um indicador de valores faltantes e usar as técnicas de indexação tradicionais. Embora, esta técnica seja útil para os métodos de indexação multidimensionais, é impraticável para os métodos de acesso métricos. Esta dissertação apresenta os resultados da pesquisa realizada para identificar e lidar com os problemas de indexação e recuperação de dados em espaços métricos com valores faltantes. Uma análise experimental dos MAMs aplicados a conjuntos de dados incompletos identificou dois problemas principais: distorção na estrutura interna do índice quando a falta é aleatória e busca tendenciosa na estrutura do índice quando o processo de falta não é aleatório. Uma variante do MAM Slim-tree, chamada Hollow-tree foi proposta com base nestes resultados. A Hollow-tree usa novas técnicas de indexação e de recuperação de dados com valores faltantes quando o processo de falta é aleatório. A técnica de indexação inclui um conjunto de políticas de indexação que visam a evitar distorções na estrutura interna dos índices. A técnica de recuperação de dados melhora o desempenho das consultas por similaridade sobre bases de dados incompletas. Essas técnicas utilizam o conceito de dimensão fractal do conjunto de dados e a densidade local da região de busca para estimar um raio de busca ideal para obter uma resposta mais correta, considerando os dados com valores faltantes como uma resposta potencial. As técnicas propostas foram avaliadas sobre diversos conjuntos de dados reais e sintéticos. Os resultados mostram que a Hollow-tree atinge quase 100% de precisão e revocação para consultas por abrangência e mais de 90% para k vizinhos mais próximos, enquanto a Slim-tree rapidamente deteriora com o aumento da quantidade de valores faltantes. Tais resultados indicam que a técnica de indexação proposta ajuda a estabelecer a consistência na estrutura do índice e a técnica de busca pode ser realizada com um desempenho notável. As técnicas propostas são independentes do MAM básico usado e podem ser aplicadas em uma grande variedade deles, permitindo estender a classe dos MAMs em geral para tratar dados faltantes.

A complexidade é uma característica de grande importância em processos de reconhecimento de padrões, especialmente naqueles que envolvem imagens biológicas. Este trabalho tem como objetivo estudar métodos que realizam a análise de imagens por meio da análise de sua complexidade. Os métodos a serem estudados foram selecionados com base na similaridade de seus algoritmos e metodologia: dimensão fractal, Caminhada Determinística do Turista e Redes Complexas. Estes métodos permitem realizar a análise e segmentação de formas ou texturas contidas em uma imagem com base na sua variação de complexidade. Dos três métodos considerados, dois deles fazem parte do estado da arte em análise de complexidade, enquanto que a dimensão fractal já é aplicada a mais tempo na análise de formas e texturas. Os trabalhos aqui desenvolvidos visam comparar e analisar os métodos selecionados por meio de experimentos com imagens de forma e texturas, sendo utilizadas texturas naturais e de Brodatz, freqüentemente utilizadas na literatura como benchmark para texturas. Com base no conhecimento adquirido, novas técnicas voltadas para a análise e segmentação de formas e texturas foram desenvolvidas, assim como foram analisadas as deficiências e propostas melhorias às técnicas estudadas. Além disso, diversos experimentos com estas metodologias foram realizados em aplicações de Bioinformática / Complexity is a feature of great importance in pattern recognition processes, especially those involving biological images. This work aims to study methods that perform image analysis by the analysis of its complexity. The methods to be studied were selected based on similarity of their algorithms and methodology: fractal dimension, Deterministic Tourist Walk and Complex Networks. These methods enable us to perform the analysis and segmentation of shapes and textures contained in an image based on the variation of its complexity. Of the three methods considered, two of them are part of the state of the art in complexity analysis, while the fractal dimension is already applied in shapes and textures analysis. The work developed here aims to compare and analyze the selected methods through experiments with shape and texture images, utilizing for this natural and Brodatz textures samples, often used in literature as benchmark for textures analysis. Based on the knowledge acquired, new techniques for analysis and segmentation of shapes and textures were developed, as also were analyzed the deficiencies and proposed improvements to the techniques studied. Moreover, several experiments with these methods were performed in bioinformatics applications

Submitted by Adriana Martinez (amartinez@unoeste.br) on 2018-02-20T17:29:34Z No. of bitstreams: 1 Gabriela Haro de Melo.pdf: 1119579 bytes, checksum: f99d268fab2103015abb1dca2740c64b (MD5) / Made available in DSpace on 2018-02-20T17:29:34Z (GMT). No. of bitstreams: 1 Gabriela Haro de Melo.pdf: 1119579 bytes, checksum: f99d268fab2103015abb1dca2740c64b (MD5) Previous issue date: 2017-09-18 / Due to its high concentration of growth factors, platelet-rich plasma (PRP) has become the subject of much research, either as an isolated or adjuvant treatment for wound repair. The fractal dimension has been used as an innovative method for the characterization and quantification of tissue structural alterations without any interference from the evaluator. The objective of this study was characterizing collagen types I and III in experimentally induced wounds and treated with different PRP sources by means of fractal dimension and other histological analyzes using blades stained with picrocirius-red and hematoxylin-eosin (HE). In statistical analysis, the variables presented a non-parametric distribution; thus, Mann-Whitney and Kruskall-Wallis tests (p<0,05) were performed using the Student-Newman-Keuls method. It was observed an increase of the collagen fibers by measuring the area with Threshold Color in the different treatments, especially with autologous PRP (p = 0.009) and heterologous PRP (p = 0.016). The fractal dimension of type I collagen had a significant increase for the autologous group (p = 0.028) in relation to the control, but, on average (p <0.05), type III collagen was higher than type I. Fractal dimension was efficient to separately characterize type I and type III collagen fibers. In this study, collagenization was higher in autologous, heterologous and homologous PRP treatments. / Devido a alta concentração em fatores de crescimento, o plasma rico em plaquetas (PRP) tornou-se objeto de estudos para pesquisadores de todo o mundo, seja como tratamento isolado ou adjuvante e demonstra efeitos benéficos em processos de cicatrização com maior produção de colágeno. Contudo, são escassos estudos que avalim a qualidade do tecido cicatricial em feridas tratadas com diferentes tipos de PRP e que qualificam o colágeno, proteína importante na elasticidade e resistência da pele. A dimensão fractal (DF), método utilizado em lâminas histológicas, permite a caracterização de estruturas. Geralmente, quanto maior a DF, mais complexa a forma do objeto e dentre os vários cálculos de DF existente, a técnica de box-couting é a mais indicada. Levantamos a hipótese de que a DF, como método auxiliar na avaliação da fisiopatologia da reparação tecidual, pode contribuir na tomada de decisão do protocolo terapêutico mais indicado para cicatrização de feridas. Objetivou-se Avaliar as características do colágeno na reparação tecidual de feridas experimentalmente induzidas e tratadas com diferentes fontes de plasma rico em plaquetas (PRP) gel por meio de achados histológicos e análise fractal. Para avaliar os efeitos de diferentes tipos de PRP sob a forma de gel em cicatrização cutânea in vivo, foram utilizados 18 coelhos Nova Zelândia, divididos em três grupos de seis animais nos quais foram induzidas duas feridas com punch de 8 mm, o lado esquerdo (controle) foi tratado com NaCl 0,9%, e o direito (tratado) foi aplicado gel de PRP. As feridas foram avaliadas por um período de 17 dias. Ao 17º dia, foram realizadas biópsias para avaliação histológica pelo software Image J®. Seis animais foram doadores para PRP homólogo e um cão adulto foi usado para obtenção do PRP heterólogo. Observou-se incremento das fibras colágenas, por meio de medição de área com Threshold Colour nos diferentes tratamentos, sobretudo com PRP de fontes autóloga (p=0,009) e homóloga (p=0,016). A DF do colágeno tipo I mostrou aumento significativo para o grupo autólogo (p=0,028) em relação ao controle, porém, em média (p<0,05) o colágeno tipo III foi maior em relação ao tipo I. A coloração de picrosirius sob luz polarizada permitiu a visualização do colágeno tipo I mostrando marcada birrefringência laranja e vermelha e do colágeno tipo III com birrefringência verde. Nos grupos tratados, as fibras colágenas do tipo I, foram mais evidentes e organizadas do que as do tipo III, que estavam distribuídas irregularmente, quando comparadas com as do tipo I. Em microscopia optica, em lâminas coradas com HE, não foram observados indicativos de necrose ou inflamação nos tecidos analisados, assim como em outros estudos. A dimensão fractal foi eficaz para caracterizar separadamente as fibras de colágeno tipo I e tipo III. Os métodos utilizados neste estudo demonstraram que o emprego de PRP gel de diferentes fontes para o tratamento de feridas dérmicas é eficiente, porém o gel autólogo demonstrou melhor colagenização quando comparado ao heterólogo e homólogo, respectivamente, homólogo e heterólogo, respectivamente.

In dieser Arbeit werden Zufallsfeld-Ising-Modelle mit einem eingefrorenen dichotomen symmetrischen Zufallsfeld für den eindimensionalen Fall und das Bethe-Gitter untersucht. Dabei wird die kanonische Zustandssumme zu der eines einzelnen Spins in einem effektiven Feld umformuliert. Im ersten Teil der Arbeit werden das mulktifraktale Spektrum dieses effektiven Feldes untersucht, Übergänge im Spektrum erklärt und Ungleichungen zwischen lokalen und globalen Dimensionsbegriffen bewiesen, die eine weitgehend vollständige Charakterisierung des multifraktalen Spektrums durch eine Reihe von Schranken erlauben. Ein weiterer Teil der Arbeit beschäftigt sich mit einer ähnlichen Charakterisierung des Maßes der lokalen Magnetisierung, das aus dem Maß des effektiven Feldes durch Faltung hervorgeht. In diesem Zusammenhang wird die Faltung von Multifraktalen in einem allgemeineren Rahmen behandelt und Zusammenhänge zwischen den multifraktalen Eigenschaften der Faltung und denen der gefalteten Maße bewiesen. Im dritten Teil der Dissertation wird der Phasenübergang von Ferro- zu Paramagnetismus im Modell auf dem Bethe Gitter untersucht. Neben verbesserten exakten Schranken für die Eindeutigkeit des paramagnetischen Zustands werden im wesentlichen drei Kriterien für die tatsächliche Lage des Übergangs angegeben und numerisch ausgewertet. Die multifraktalen Eigenschaften des effektiven Felds im Modell auf dem Bethe-Gitter schließlich erweisen sich als trivial, da die interessanten Dimensionen nicht existieren. / In this work random field Ising models with quenched dichotomous symmetric random field are considered for the one-dimensional case and on the Bethe lattice. To this end the canonical partition function is reformulated to the partition function of one spin in an effective field. In the first part of the work the multifractal spectrum of this effective field is investigated, transitions in the spectrum are explained and inequalities between local and global generalized fractal dimensions are proven which allow to characterize the multifractal spectrum bei various bounds. A further part of the work is dedicated to the characterization of the measure of the local magnetization which is obtained by convolution of the measure of the effective field with itself. In this context the convolution of multifractals is investigated in a more general setup and relations between the multifractal properties of the convolution and the multifractal properties of the convoluted measures are proven. The phase transition from ferro- to paramagnetismus for the model on the Bethe lattice is investigated in the third part of the thesis. Apart from improved exact bounds for the uniqueness of the paramagnetic state essentially three criteria for the transition are developped and numerically evaluated to determine the transition line. The multifractal properties of the effective field for the model on the Bethe lattice finally turn out to be trivial because the interesting dimensions do not exist.

Multimodal biometric systems have been widely applied in many real-world applications due to its ability to deal with a number of significant limitations of unimodal biometric systems, including sensitivity to noise, population coverage, intra-class variability, non-universality, and vulnerability to spoofing. This PhD thesis is focused on the combination of both the face and the left and right irises, in a unified hybrid multimodal biometric identification system using different fusion approaches at the score and rank level. Firstly, the facial features are extracted using a novel multimodal local feature extraction approach, termed as the Curvelet-Fractal approach, which based on merging the advantages of the Curvelet transform with Fractal dimension. Secondly, a novel framework based on merging the advantages of the local handcrafted feature descriptors with the deep learning approaches is proposed, Multimodal Deep Face Recognition (MDFR) framework, to address the face recognition problem in unconstrained conditions. Thirdly, an efficient deep learning system is employed, termed as IrisConvNet, whose architecture is based on a combination of Convolutional Neural Network (CNN) and Softmax classifier to extract discriminative features from an iris image. Finally, The performance of the unimodal and multimodal systems has been evaluated by conducting a number of extensive experiments on large-scale unimodal databases: FERET, CAS-PEAL-R1, LFW, CASIA-Iris-V1, CASIA-Iris-V3 Interval, MMU1 and IITD and MMU1, and SDUMLA-HMT multimodal dataset. The results obtained have demonstrated the superiority of the proposed systems compared to the previous works by achieving new state-of-the-art recognition rates on all the employed datasets with less time required to recognize the person’s identity.Multimodal biometric systems have been widely applied in many real-world applications due to its ability to deal with a number of significant limitations of unimodal biometric systems, including sensitivity to noise, population coverage, intra-class variability, non-universality, and vulnerability to spoofing. This PhD thesis is focused on the combination of both the face and the left and right irises, in a unified hybrid multimodal biometric identification system using different fusion approaches at the score and rank level. Firstly, the facial features are extracted using a novel multimodal local feature extraction approach, termed as the Curvelet-Fractal approach, which based on merging the advantages of the Curvelet transform with Fractal dimension. Secondly, a novel framework based on merging the advantages of the local handcrafted feature descriptors with the deep learning approaches is proposed, Multimodal Deep Face Recognition (MDFR) framework, to address the face recognition problem in unconstrained conditions. Thirdly, an efficient deep learning system is employed, termed as IrisConvNet, whose architecture is based on a combination of Convolutional Neural Network (CNN) and Softmax classifier to extract discriminative features from an iris image. Finally, The performance of the unimodal and multimodal systems has been evaluated by conducting a number of extensive experiments on large-scale unimodal databases: FERET, CAS-PEAL-R1, LFW, CASIA-Iris-V1, CASIA-Iris-V3 Interval, MMU1 and IITD and MMU1, and SDUMLA-HMT multimodal dataset. The results obtained have demonstrated the superiority of the proposed systems compared to the previous works by achieving new state-of-the-art recognition rates on all the employed datasets with less time required to recognize the person’s identity. / Higher Committee for Education Development in Iraq

No description available.

La sédimentation des suspensions aqueuses de particules minérales microniques, polydisperses et concentrées a été analysée à l’aide du Turbiscan MA 2000 fondé sur la diffusion multiple de la lumière, en vue d’établir la procédure qui permet de déceler la présence d’une morphologie fractale, puis de déduire les règles de comportements des suspensions fractales par la modélisation avec les réseaux de neurones. Le domaine des interactions interparticulaires physicochimiques (0 à 10% volumique en solide) a été privilégié.La méthodologie de détermination de la structure multifractale des agglomérats et de la suspension a été proposée. La modification structurale des agglomérats qui est à l’origine de comportements non linéaires des suspensions et qui dépend des propriétés cohésives des particules primaires, est interprétée par la variation de la mobilité électrophorétique des particules en suspension. Une approche d’estimation de ces modifications structurales par les réseaux de neurones, à travers la dimension fractale, a été présentée. Les limites du modèle à assimiler ces comportements particuliers ont été expliquées comme résultant du faible nombre d’exemples et de la grande variabilité des mesures aux faibles fractions volumiques en solide. / The sedimentation of aqueous suspensions of micron-sized mineral particles, polydisperses and concentrated, was analyzed using the Turbiscan MA 2000 based on the multiple light scattering in order to establish the procedure to detect the presence of a fractal morphology, and then to deduce the set of laws of fractal behavior of suspensions by modeling with neural networks. The methodology for determining the multifractal structure of agglomerates and the suspension was proposed. The structural modifications of the agglomerates at the origin of the nonlinear behavior of suspensions and which depends on cohesive properties of primary particles, is interpreted by the change of the electrophoretic mobility of suspended particles. The estimation by neural networks of these structural changes, through the fractal dimension has been presented. The limits of the model to learn these specific behaviors have been explained as resulting from the low number of examples and the great variability in the measurements at low volume fractions of solid.