La reconnaissance faciale est une technologie importante en vision par ordinateur, avec un rôle central en biométrie, interface homme-machine, contrôle d’accès, indexation multimédia, etc. L’occultation partielle, qui change complétement l’apparence d’une partie du visage, ne provoque pas uniquement une dégradation des performances en reconnaissance faciale, mai peut aussi avoir des conséquences en termes de sécurité. Dans cette thèse, nous concentrons sur le problème des occultations en reconnaissance faciale en environnements non contrôlés. Nous proposons une séquence qui consiste à analyser de manière explicite les occultations et à fiabiliser la reconnaissance faciale soumises à diverses occultations. Nous montrons dans cette thèse que l’approche proposée est plus efficace que les méthodes de l’état de l’art opérant sans traitement explicite dédié aux occultations. Nous identifions deux nouveaux types d’occultations, à savoir éparses et dynamiques. Des solutions sont introduites pour gérer ces problèmes d’occultation nouvellement identifiés dans un contexte de vidéo surveillance avancé. Récemment, le nouveau capteur Kinect a été utilisé avec succès dans de nombreuses applications en vision par ordinateur. Nous introduisons ce nouveau capteur dans le contexte de la reconnaissance faciale, en particulier en présence d’occultations, et démontrons son efficacité par rapport aux caméras traditionnelles. Finalement, nous proposons deux approches basées 2D et 3D permettant d’améliorer les techniques de base en reconnaissance de visages. L’amélioration des méthodes de base peut alors générer un impact positif sur les résultats de reconnaissance en présence d’occultations. / Face recognition is an important technology in computer vision, which often acts as an essential component in biometrics systems, HCI systems, access control systems, multimedia indexing applications, etc. Partial occlusion, which significantly changes the appearance of part of a face, cannot only cause large performance deterioration of face recognition, but also can cause severe security issues. In this thesis, we focus on the occlusion problem in automatic face recognition in non-controlled environments. Toward this goal, we propose a framework that consists of applying explicit occlusion analysis and processing to improve face recognition under different occlusion conditions. We demonstrate in this thesis that the proposed framework is more efficient than the methods based on non-explicit occlusion treatments from the literature. We identify two new types of facial occlusions, namely the sparse occlusion and dynamic occlusion. Solutions are presented to handle the identified occlusion problems in more advanced surveillance context. Recently, the emerging Kinect sensor has been successfully applied in many computer vision fields. We introduce this new sensor in the context of face recognition, particularly in presence of occlusions, and demonstrate its efficiency compared with traditional 2D cameras. Finally, we propose two approaches based on 2D and 3D to improve the baseline face recognition techniques. Improving the baseline methods can also have the positive impact on the recognition results when partial occlusion occurs.

Made available in DSpace on 2014-06-11T19:22:25Z (GMT). No. of bitstreams: 0 Previous issue date: 2007-07-31Bitstream added on 2014-06-13T18:08:14Z : No. of bitstreams: 1 holzschuh_ml_me_prud.pdf: 1897806 bytes, checksum: 49dcd938d6b1fba152b414f9f4c978cc (MD5) / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) / Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP) / A ocultação, um fenômeno pelo qual um corpo celeste deixa de ser visto em razão da sobreposição de um outro, começou a ser usada para estudar a atmosfera de outros planetas há vários anos. Com o advento dos satélites artificiais GPS (Global Positioning System), em conjunto com os satélites LEO (Low Earth Orbiting), surge a Rádio Ocultação GPS (ROGPS), abrindo-se novas perspectivas para as pesquisas sobre o clima e previsão de tempo. Na RO GPS, os sinais que são enviados pelo satélite GPS e recebidos no satélite LEO sofrem os efeitos da troposfera e da ionosfera ao atravessarem a atmosfera terrestre, afetando a fase e a amplitude das ondas dos sinais envolvidos. A propagação do sinal sofre refração formando uma linha curva entre o satélite transmissor e o receptor, o que permite calcular diferentes índices de refratividade, os quais carregam informações da composição da atmosfera terrestre. A partir de perfis desses índices de refratividade é possível extrair perfis de temperatura, umidade e pressão atmosférica na troposfera, além da distribuição na densidade de elétrons na ionosfera terrestre. Vários trabalhos foram realizados para verificar a qualidade dos perfis atmosféricos obtidos com o emprego do satélite CHAMP (Challenging Minisatellite Payload of Geophysical Research and Application), um satélite LEO, em diversas regiões do globo. Nesse trabalho, objetiva-se investigar e implementar a técnica de ocultação GPS, bem como avaliar a qualidade de perfis de temperatura obtidos com o satélite CHAMP sobre a América do Sul, comparando-os com perfis de radiossondas e perfis provenientes da condição inicial do modelo de Previsão Numérica de Tempo (PNT). A implementação da recuperação de um perfil da atmosfera foi realizada utilizando o código C/A para o cálculo do ângulo de curvatura e o parâmetro de impacto, cujos valores aproximados foram 0,043° e 6381 km, respectivamente. / The occultation, phenomenon in which a celestial body in not seen because of the superposition of another one, started being used several years ago to study the atmosphere of other planets. With the advent of GPS (Global Positioning System) satellites, together with the LEO (Low Earth Orbit) satellites, the GPS Radio Occultation (GPS-RO) arises, in order to obtain new perspectives in studies related to climate and weather forecast. In GPS-RO, the signals sent by the GPS satellite and received by the LEO satellite are disturbed by the ionosphere and troposphere while cross the terrestrial atmosphere, this affects the phase and the amplitude from the involved signals. The signals propagation is affected by refraction, this generates a bending angle between the transmitter and receiver satellites, allowing computing different refractivity index, which carry information about the terrestrial atmosphere composition. Using these profiles of refractivity index is possible to extract profiles of temperature, humidity, atmospheric pressure in the troposphere, and to obtain the electronic density distribution in terrestrial ionosphere. Several works were realized to verify the quality of atmospheric profiles obtained using the CHAMP (Challenging Minisatellite Payload of Geophysical Research and Application) satellite, LEO satellite, in several regions of the globe. In this research the goal is to investigate and implement the GPS occultation technique, and additionally to evaluate the quality of the temperature profiles obtained by CHAMP satellite over the South of America, comparing them with the RSO (Radiosonde) profiles and with the profiles obtained by the initial condition of Numeric Weather Prediction (NWP) model.

Measurements made by the Voyager 2 spacecraft during its flyby of Uranus in 1986 found warm stratospheric and hot thermospheric temperatures that cannot be explained by solar heating alone. This contributes to what has become known as the “giant planet energy crisis”: there is a fundamental lack of understanding of the energy balance of the giant planets in the solar system. Uranus has the coldest stratospheric temperatures, the hottest thermospheric temperatures at some altitudes, and yet the weakest internal heat flux of all four giant planets. Moreover, the Voyager 2 temperature profiles are at odds with the many contemporaneous Earth-based stellar occultation observations. This unresolved tension impedes efforts to compare atmospheres in the solar system to one another and to exoplanet atmospheres. In this dissertation, I present an investigation into the upper atmospheric temperatures of Uranus using archival Earth-based stellar occultation observations. I begin with an overview of planetary atmospheres and remote-sensing measurements of Uranus in Chapter 1. In Chapter 2, I derive and explain how stellar occultations are predicted, observed, processed, and analyzed, emphasizing my contributions. Chapter 3 describes how I validated these techniques on an archival Mars occultation. In Chapter 4, I present the results of comparing the Voyager 2 measurements to 26 archival Earth-based stellar occultations by Uranus. In Chapter 5, I present new temperature profiles from reprocessing these 26 occultations and a new one-dimensional atmospheric model based thereon. Chapter 6 outlines a low-Earth orbit mission concept to observe many new stellar occultations. Chapter 7 contains conclusions and summaries. My primary finding is that the lower thermosphere of Uranus is much cooler than reported by Voyager 2. I find that the mesopause is likely higher in altitude than previously believed and the stratosphere of Uranus has a nearly isothermal section, in alignment with the other giant planets. My new atmospheric model suggests Uranus has a significant source of heat dissipation in the stratosphere, which might be supplied by gravity waves. This work can contribute to planning for any Uranus Orbiter and Probe mission by helping to revise Uranus’ representative temperature profile and improving our understanding of Uranus’ energy balance.

Mémoire numérisé par la Direction des bibliothèques de l'Université de Montréal.

When a radio signal traverses the atmosphere it will be delayed by not only thedistance between transmitter and receiver, but also the atmosphere. Given knowl-edge of the characteristics of the sent signal the effect of the atmosphere can beobtained from the received signal. This concept is called radio occultation. Radiooccultation can provide high accuracy profiles of temperature, pressure and watervapour troughout the atmosphere.This report aims to present the work and results from a thesis performed atRUAG Space in Göteborg. The purpose of the thesis was to implement a simulatorwhich with high accuracy could generate a signal as it would have been receivedhad it propagated through the atmosphere.We will show that the generated signal passes the requirements that have beenset. / När en radiovåg passerar genom atmosfären kommer den att fördröjas, inte bara avavståndet mellan sändare och mottagare utan också av atmosfären. Givet kunskapom karaktäristiken hos den sända signalen kan atmosfärens effekt erhållas från denmottagna signalen. Detta koncept kallas för radio-ockultation. Radio-ockultationkan med hög noggrannhet ge profiler för temperatur, tryck och vattenånga genomatmosfären.Denna rapport ämnar presentera det jobb och de resultat som uppnåtts genomett examensarbete genomfört på RUAG Space i Göteborg. Examensarbetets syftevar att implementera en simulator som med hög noggrannhet kan generera ensignal så som den hade sett ut då den propagerat genom atmosfären.Vi kommer att visa att den genererade signalen uppnår de krav som ställts.

Resumo: A ocultação, um fenômeno pelo qual um corpo celeste deixa de ser visto em razão da sobreposição de um outro, começou a ser usada para estudar a atmosfera de outros planetas há vários anos. Com o advento dos satélites artificiais GPS (Global Positioning System), em conjunto com os satélites LEO (Low Earth Orbiting), surge a Rádio Ocultação GPS (ROGPS), abrindo-se novas perspectivas para as pesquisas sobre o clima e previsão de tempo. Na RO GPS, os sinais que são enviados pelo satélite GPS e recebidos no satélite LEO sofrem os efeitos da troposfera e da ionosfera ao atravessarem a atmosfera terrestre, afetando a fase e a amplitude das ondas dos sinais envolvidos. A propagação do sinal sofre refração formando uma linha curva entre o satélite transmissor e o receptor, o que permite calcular diferentes índices de refratividade, os quais carregam informações da composição da atmosfera terrestre. A partir de perfis desses índices de refratividade é possível extrair perfis de temperatura, umidade e pressão atmosférica na troposfera, além da distribuição na densidade de elétrons na ionosfera terrestre. Vários trabalhos foram realizados para verificar a qualidade dos perfis atmosféricos obtidos com o emprego do satélite CHAMP (Challenging Minisatellite Payload of Geophysical Research and Application), um satélite LEO, em diversas regiões do globo. Nesse trabalho, objetiva-se investigar e implementar a técnica de ocultação GPS, bem como avaliar a qualidade de perfis de temperatura obtidos com o satélite CHAMP sobre a América do Sul, comparando-os com perfis de radiossondas e perfis provenientes da condição inicial do modelo de Previsão Numérica de Tempo (PNT). A implementação da recuperação de um perfil da atmosfera foi realizada utilizando o código C/A para o cálculo do ângulo de curvatura e o parâmetro de impacto, cujos valores aproximados foram 0,043° e 6381 km, respectivamente. / Abstract: The occultation, phenomenon in which a celestial body in not seen because of the superposition of another one, started being used several years ago to study the atmosphere of other planets. With the advent of GPS (Global Positioning System) satellites, together with the LEO (Low Earth Orbit) satellites, the GPS Radio Occultation (GPS-RO) arises, in order to obtain new perspectives in studies related to climate and weather forecast. In GPS-RO, the signals sent by the GPS satellite and received by the LEO satellite are disturbed by the ionosphere and troposphere while cross the terrestrial atmosphere, this affects the phase and the amplitude from the involved signals. The signals propagation is affected by refraction, this generates a bending angle between the transmitter and receiver satellites, allowing computing different refractivity index, which carry information about the terrestrial atmosphere composition. Using these profiles of refractivity index is possible to extract profiles of temperature, humidity, atmospheric pressure in the troposphere, and to obtain the electronic density distribution in terrestrial ionosphere. Several works were realized to verify the quality of atmospheric profiles obtained using the CHAMP (Challenging Minisatellite Payload of Geophysical Research and Application) satellite, LEO satellite, in several regions of the globe. In this research the goal is to investigate and implement the GPS occultation technique, and additionally to evaluate the quality of the temperature profiles obtained by CHAMP satellite over the South of America, comparing them with the RSO (Radiosonde) profiles and with the profiles obtained by the initial condition of Numeric Weather Prediction (NWP) model. / Orientador: João Francisco Galera Monico / Coorientador: Luiz Fernando Sapucci / Banca: Hisao Takahashi / Banca: José Tadeu Garcia Tommaselli / Mestre

The innermost Galilean satellite, Io, is the most volcanically active body in the solar system. The volcanic activity leads to material being released into Jupiter's magnetosphere near Io's orbit. This material becomes ionized and trapped in Jupiter's rotating magnetic field. The trapped material forms a torus of material around Jupiter that is called the Io plasma torus. It contains an inner cold torus and an outer warm torus. In this dissertation, I determine and interpret the distribution of plasma in the Io plasma torus from radio occultation observations by the Juno spacecraft. I perform a feasibility study to show that Juno radio occultation observations should be able to detect the Io plasma torus. Based on this feasibility study, I predict that key Io plasma torus parameters -- value and location of maximum total electron content, and scale height -- can be determined with 10--20 percent uncertainties. I analyze Juno radio occultation observations from Perijove 1. Perijove is the point where the spacecraft makes its closest approach. The observations are taken for a 6 hour period around perijove and are labeled as Perijove followed by a number which corresponds to the orbit number. From the observations I determine the Io plasma torus parameters and find that inferred densities are ~30% larger than models suggested. These results show that Juno radio occultation observations can detect and usefully characterize the Io plasma torus. I analyze data from Perijoves 3, 6, and 8 and determine how Io plasma torus parameters vary. In this set of observations, the warm torus maximum total electron content and scale height do not vary greatly. I test the prediction that the torus lies in the centrifugal equator by modeling the equator location. Observed and predicted locations agree reasonably if a Juno magnetic field model and a simple current sheet model are used. I find that the contribution of the current sheet is significant, which suggests that remote observations of the location of the Io plasma torus can be used to constrain Jupiter's magnetospheric current sheet.

CubeSats provide an inexpensive means for space-based research. However, optimal mission design depends on minimizing payload size and power. This thesis investigates the GLO (GFCR (Gas Filter Correlation Radiometry) Limb Occultation) prototype, a new small-form-factor design that enables sub-kilometer resolution of the vertical profile of atmospheric trace species to determine radiative influences. This technology improves SWAP (Size, Weight, And Power) over heritage SOFIE and HALOE instruments and provides a cost-effective alternative for solar occultation limb monitoring. A python script was developed to analyze solar intensity through GLO telescope channels. Non-uniform aerosol images used a peak intensity algorithm compared to the edge detection function designed for GFCR channels. Scaling corrections were made for beam splitter inaccuracy and SNR was characterized for frame collection. Different cameras were tested to weigh accuracy versus cost of a camera baffle. Using the Langley plot method, solar intensity versus changes in the solar zenith angle were measured for extrapolation of optical depths. AERONET, a network of ground-based sun photometers measuring atmospheric aerosols, was used for aerosol optical depth validation. Spectral Calculator transmission data allowed for GFCR vacuum channel comparison, gas cell spectral analysis, and gas cell to vacuum channel optical depth examination. Ground testing provided promising results with the low-cost prototype. It will be further evaluated through a balloon flight demonstration using a flight-ready GLO instrument. Additionally, analysis for the DUSTIE mission is planned and simulated using STK and Matlab. This includes CubeSat bus selection, orbit analysis for occultation occurrences, power budgeting, and communication capabilities. / Master of Science / Cube Satellites (CubeSats) provide an inexpensive means for space-based research. However, optimal mission design depends on minimizing payload size and power. This thesis investigates the GLO (GFCR (Gas Filter Correlation Radiometry) Limb Occultation) prototype. This technology will determine the influences on the energy balance between the Earth and atmosphere due to aerosol and gas particle concentrations. This is implemented with improved SWAP (Size, Weight, And Power) compared to previously flown instruments. Scaling corrections were made for beam splitter inaccuracy and the Signal-to-Noise Ratio (SNR) was characterized for frame collection for the demonstration GLO instrument. The changing solar intensity as the sun moved across the sky was measured to infer aerosol and gas concentrations in the atmosphere. A network of ground-based sun photometers measuring atmospheric aerosols was used to validate aerosol concentration measurements. GLO vacuum channel measurements and gas cell properties were compared to transmission simulations for accuracy. Ground testing provided promising results with the low-cost prototype. It will be further evaluated through a balloon flight demonstration using a flight-ready GLO instrument. Additionally, analysis for the Dust Sounder and Temperature Imager Experiment (DUSTIE) mission is planned and simulated using STK and Matlab. This includes CubeSat bus selection, orbit analysis for occultation occurrences, power budgeting, and communication capabilities.

Ce projet de thèse porte sur l'étude de la haute atmosphère de Titan à partir de mesures en laboratoire de sections efficaces d'absorption et de l'analyse des données de Cassini-UVIS.La caractérisation de l'instrument et des observations effectuées par UVIS était nécessaire. Les données provenant des canaux ultraviolet lointain (FUV) et ultraviolet extrême (EUV) d'UVIS ont été analysées et corrigées des effets instrumentaux. A partir de l'analyse de huit occultations solaires dans l'EUV, les profils de densité volumique de N2 et CH4 ont été déterminés par une méthode d'inversion avec régularisation. Les températures ont ensuite été obtenues à partir des profils de N2 en supposant une haute atmosphère isotherme. Les occultations stellaires dans le FUV ont été modélisés et une technique de détermination des densités caractérisée. La possibilité de détection de différentes molécules (dont certaines n'avaient jamais été observés par cette technique avant) a été analysée. Puis, en utilisant l'algorithme de minimisation de Levenberg–Marquardt, les profils de densité de colonne pour différents hydrocarbures et de profondeurs optiques pour les aérosols ont été obtenus à partir de données simulés. Les densités de colonne ont été inversées avec une procédure de régularisation afin d'obtenir des profils de densité volumique pour les hydrocarbures et des profils d'extinction pour les aérosols. La procédure a finalement été appliquée à deux occultations stellaires mesurées par UVIS. Les hydrocarbures étudiés sont CH4, C2H2, HCN, C2H4, C4H2, HC3N et C6H6.Les profils issus des occultations stellaires et solaires ont été obtenus pour différentes dates et des différents lieux. Les profils et les températures dérivés ont donc été analysés en fonction des variables géographiques et temporelles : latitude, longitude, date d'observation, etc. La variabilité atmosphérique est discutée à la lumière de ces résultats.Le benzène (C6H6), une molécule détectée dans l'atmosphère de Titan, est particulièrement importante car elle est considéré comme intermédiaire entre le gaz et la formation des particules solides. Des mesures de l'absorption du benzène dans le domaine ultraviolet, à des températures qui couvrent une gamme de température allant de l'ambiante à 215 K, ont été réalisées dans différentes installations internationales. A partir de ces mesures, la section efficace d'absorption du benzène a été déterminée et analysée en termes des transitions observées et en fonction de la température de mesure. Ces résultats ont été utilisés dans le calcul d'abondance de C6H6 dans la thermosphère de Titan comme indiqué dans les paragraphes précédents.En résumé, l'analyse des observations UVIS présentées contribuent à la caractérisation de la haute atmosphère à travers des profils de N2, de températures thermosphériques et des profils de différents hydrocarbures et nitrile. Ces données d'observation aideront à contraindre les modèles photochimiques. Les profils d'abondance donnés en fonction de différentes coordonnées géographiques et temporelles pourront être utilisés pour étudier plus avant la variabilité atmosphérique. Les résultats de ce travail aideront donc à la compréhension de la composition et la dynamique de l'haute atmosphère de Titan. Cette connaissance, combinée avec des informations sur la basse atmosphère et la surface de Titan, aidera à comprendre l'évolution de molécules organiques dans ce corps céleste abiotique voisin.Les résultats de ce travail aideront donc à la compréhension de la composition et la dynamique de la haute atmosphère de Titan. Cette connaissance, combinée avec des informations sur la basse atmosphère et la surface de Titan, aidera à comprendre l'évolution de molécules organiques dans ce corps céleste abiotique voisin / This PhD project focuses on the study of the upper atmosphere of Titan from the analysis of Cassini-UVIS data and laboratory measurements of absorption cross sections.A characterization of the UVIS instrument and observations was necessary. Data from the Far UltraViolet (FUV) and Extreme UltraViolet (EUV) channels of UVIS were analyzed and corrected for instrument effects. From the analysis of 8 solar occultations in EUV, N2 and CH4 number density profiles were derived with an inversion regularization method. Temperatures were obtained from the N2 profiles assuming an isothermal upper atmosphere. Stellar occultations in FUV were modeled and a density retrieval technique characterized. The possibility of detection for different molecules (some of them not detected by this technique before) was analyzed. Then, using a Levenberg-Marquardt minimization algorithm, column density profiles for different hydrocarbons and nitriles, and optical depth of aerosols were obtained from simulated data. The column densities and optical depth were inverted with a regularization method to obtain number density profiles for the molecules and extinction profiles for the aerosols. The procedure was finally applied to 2 stellar occultations measured by UVIS. The species studied are CH4, C2H2, HCN, C2H4, C4H2, C6H6, HC3N, CH3, and aerosols (AER). The profiles from the stellar and solar occultations were obtained for different times and locations. The temperatures derived were analyzed as a function of geographical and temporal variables---latitude, longitude, date of observation, etc.---without a clear correlation with any of them, although a trend of decreasing temperature towards the poles could be observed. The globally averaged temperature obtained is (144 +/- 2) K. Atmospheric variability was discussed on the light of these results.Benzene (C6H6) is an important molecule detected in Titan's atmosphere because it is thought to be intermediate between the gas and solid particle formation. Measurements of absorption in the ultraviolet by benzene gas, at temperatures covering the range from room temperature to 215 K, were performed in different international facilities. From them, benzene absorption cross sections were derived and analyzed in terms on the transitions observed. No significant variation with measurement temperature was observed. Implications of this results for the identification of benzene in Titan's thermosphere by UVIS were discussed. The absorption cross sections were used in the derivation of C6H6 abundances in Titan's thermosphere commented above.In summary the analysis of UVIS observations presented contribute to the characterization of the upper atmosphere through N2 density profiles, thermospheric temperatures, density profiles of minor species and extinction profiles from aerosols. This observational data will help to constrain and contrast photochemical models. The abundance profiles and temperatures given for different geographical and temporal coordinates can be used to further study the atmospheric variability. As a whole, the results of this work are expected to help in the understanding of Titan's upper atmospheric composition and dynamics. This knowledge, combined with information about Titan's lower atmosphere and surface, will help to understand the evolution of organic molecules in this neighboring abiotic celestial body

This thesis is executed in cooperation with RUAG Space AB, which specializes in highly reliable on-board satellite equipment. The thesis focuses on the effect, which disturbs the amplitude and phase of a Global Positioning System (GPS) signal, called scintillation effect. It has a substantial impact on a GPS signal, during Radio Occultation (RO). RO is a method of analysis of a refracted signal which passes through the atmosphere. RO can be used for measuring climate change and for weather forecasting. By retrieving the bending angle of a GPS signal, three basic parameters of the Earth’s atmosphere can be obtained at different heights: temperature, pressure and humidity. As the scintillation effect causes prominent errors in the bending angle calculations, it is crucial to provide possibly the most precise mathematical model, which allows to conceive proper ionospheric corrections. In this thesis, the model using Rytov approach is implemented and optimised with different optimisation functions. It is shown that the scintillation model can be optimized, which may contribute to a more accurate retrieval of the atmospheric profiles.