De nos jours les cartes d’utilisation/occupation du sol (USOS) à une échelle régionale sont habituellement générées à partir d’images satellitales de résolution modérée (entre 10 m et 30 m). Le National Land Cover Database aux États-Unis et le programme CORINE (Coordination of information on the environment) Land Cover en Europe, tous deux fondés sur les images LANDSAT, en sont des exemples représentatifs. Cependant ces cartes deviennent rapidement obsolètes, spécialement en environnement dynamique comme les megacités et les territoires métropolitains. Pour nombre d’applications, une mise à jour de ces cartes sur une base annuelle est requise. Depuis 2007, le USGS donne accès gratuitement à des images LANDSAT ortho-rectifiées. Des images archivées (depuis 1984) et des images acquises récemment sont disponibles. Sans aucun doute, une telle disponibilité d’images stimulera la recherche sur des méthodes et techniques rapides et efficaces pour un monitoring continue des changements des USOS à partir d’images à résolution moyenne. Cette recherche visait à évaluer le potentiel de telles images satellitales de résolution moyenne pour obtenir de l’information sur les changements des USOS à une échelle régionale dans le cas de la Communauté Métropolitaine de Montréal (CMM), une métropole nord-américaine typique. Les études précédentes ont démontré que les résultats de détection automatique des changements dépendent de plusieurs facteurs tels : 1) les caractéristiques des images (résolution spatiale, bandes spectrales, etc.); 2) la méthode même utilisée pour la détection automatique des changements; et 3) la complexité du milieu étudié. Dans le cas du milieu étudié, à l’exception du centre-ville et des artères commerciales, les utilisations du sol (industriel, commercial, résidentiel, etc.) sont bien délimitées. Ainsi cette étude s’est concentrée aux autres facteurs pouvant affecter les résultats, nommément, les caractéristiques des images et les méthodes de détection des changements. Nous avons utilisé des images TM/ETM+ de LANDSAT à 30 m de résolution spatiale et avec six bandes spectrales ainsi que des images VNIR-ASTER à 15 m de résolution spatiale et avec trois bandes spectrales afin d’évaluer l’impact des caractéristiques des images sur les résultats de détection des changements. En ce qui a trait à la méthode de détection des changements, nous avons décidé de comparer deux types de techniques automatiques : (1) techniques fournissant des informations principalement sur la localisation des changements et (2)techniques fournissant des informations à la fois sur la localisation des changements et sur les types de changement (classes « de-à »). Les principales conclusions de cette recherche sont les suivantes : Les techniques de détection de changement telles les différences d’image ou l’analyse des vecteurs de changements appliqués aux images multi-temporelles LANDSAT fournissent une image exacte des lieux où un changement est survenu d’une façon rapide et efficace. Elles peuvent donc être intégrées dans un système de monitoring continu à des fins d’évaluation rapide du volume des changements. Les cartes des changements peuvent aussi servir de guide pour l’acquisition d’images de haute résolution spatiale si l’identification détaillée du type de changement est nécessaire. Les techniques de détection de changement telles l’analyse en composantes principales et la comparaison post-classification appliquées aux images multi-temporelles LANDSAT fournissent une image relativement exacte de classes “de-à” mais à un niveau thématique très général (par exemple, bâti à espace vert et vice-versa, boisés à sol nu et vice-versa, etc.). Les images ASTER-VNIR avec une meilleure résolution spatiale mais avec moins de bandes spectrales que LANDSAT n’offrent pas un niveau thématique plus détaillé (par exemple, boisés à espace commercial ou industriel). Les résultats indiquent que la recherche future sur la détection des changements en milieu urbain devrait se concentrer aux changements du couvert végétal puisque les images à résolution moyenne sont très sensibles aux changements de ce type de couvert. Les cartes indiquant la localisation et le type des changements du couvert végétal sont en soi très utiles pour des applications comme le monitoring environnemental ou l’hydrologie urbaine. Elles peuvent aussi servir comme des indicateurs des changements de l’utilisation du sol. De techniques telles l’analyse des vecteurs de changement ou les indices de végétation son employées à cette fin. / Nowadays land use/land cover maps at regional scale are commonly generated with satellite data of medium spatial resolution (between 10 m and 30m). The National Land Cover Database (NLCD) in the United States and the Coordination of Information on the Environment (CORINE) Land Cover program in Europe, both based on LANDSAT images, are two typical examples. However, these maps become rapidly obsolete, especially in highly dynamic areas such as mega cities and metropolitan areas. In many applications, such as to monitor the water quality affected by the Land use/Land cover (LULC) change, the spread of invasive species, policy making for city managers, annual updating of LULC maps is required. Since 2007, the USGS offers access to ortho-rectified LANDSAT imagery free of charge. Both archived (since 1984) and recently acquired images are available. Without doubt, such data availability will stimulate the research on fast and cost effective methods and techniques for “continuous” regional land cover/use map updating using medium resolution satellite imagery. The objective of this research was to evaluate the potential of such medium resolution satellite imagery for providing information on changes useful for the continuous updating of LULC maps at a regional scale in the case of the Montreal Metropolitan Community (MMC) area, a typical North American metropolis. Previous studies have demonstrated that many factors could affect the results of automatic change detection such as: (1) the characteristics of the images (spatial resolution, spectral bands, etc.); (2) the method itself used to automatically detect changes; and (3) the complexity of the landscape. In the study site except for the Central Business District (CBD) and some commercial streets, land uses (industrial, commercial, residential, etc.) are well delimited. Thus this study was focused on the other factors affecting change detection results, namely, the characteristics of the images and the method of change detection. We used 6 spectral bands of LANDSAT TM/ETM+ with 30 m spatial resolution and 3 spectral bands of ASTER-VNIR with 15 m spatial resolution to evaluate the impact of image characteristics on change detection. Concerning the change detection method, we decided to compare two types of automatic techniques: (1) techniques providing information principally on the location of changed areas,and (2) techniques providing information on both the location of changed areas and the type of changes ("from-to" classes). The main conclusions of this research are as follows: Change detection techniques such as image differencing or change vector analysis applied to LANDSAT multi-temporal imagery provide an accurate picture of changed areas in a fast and efficient manner. They can thus be integrated in a continuous monitoring system for a rapid evaluation of the volume of changes. The produced maps could be helpful to guide the acquisition of high spatial resolution imagery if a detailed identification of the type of changes is required. Change detection techniques such as principal component analysis and post-classification comparison applied to LANDSAT multi-temporal imagery could provide a relatively accurate picture of “from-to” classes but at a very general thematic level (for example, built-up to green space and vice-versa, forest lands to bare soil and vice-versa, etc.). ASTER images with better spatial resolution but with less spectral bands than LANDSAT images do not provide more detailed thematic information (for example forest land to commercial or industrial areas). The results indicate that future research should be focused on the detection of changes in the vegetation cover as medium resolution imagery is highly sensitive to this type of surface cover. Maps indicating the location and the type of changes in vegetation cover are in itself very useful for various applications, such as environmental monitoring or urban hydrology, and can be used as indicators on land use changes. Techniques such as change vector analysis or vegetation indices could be used to this end.

Many mobile robot applications require robots to act safely and intelligently in complex unfamiliarenvironments with little structure and limited or unavailable human supervision. As arobot is forced to operate in an environment that it was not engineered or trained for, various aspectsof its performance will inevitably degrade. Roboticists equip robots with powerful sensorsand data sources to deal with uncertainty, only to discover that the robots are able to make onlyminimal use of this data and still find themselves in trouble. Similarly, roboticists develop andtrain their robots in representative areas, only to discover that they encounter new situations thatare not in their experience base. Small problems resulting in mildly sub-optimal performance areoften tolerable, but major failures resulting in vehicle loss or compromised human safety are not.This thesis presents a series of online algorithms to enable a mobile robot to better deal withuncertainty in unfamiliar domains in order to improve its navigational abilities, better utilizeavailable data and resources and reduce risk to the vehicle. We validate these algorithms throughextensive testing onboard large mobile robot systems and argue how such approaches can increasethe reliability and robustness of mobile robots, bringing them closer to the capabilitiesrequired for many real-world applications.

Nesta dissertação está proposta uma metodologia para fins de detecção de mudanças do uso do solo em imagens multitemporais de sensoriamento remoto. Em lugar de classificar os pixels de imagens que cobrem uma cena, em duas classes exaustivas e mutuamente excludentes (mudança, não-mudança), propõe-se adotar uma abordagem do tipo fuzzy, na qual são estimados os graus de pertinência às classes mudança e não-mudança. Com este objetivo adota-se aqui uma abordagem em nível de sub-pixel na estimação dos graus de pertinência para cada pixel. Esta abordagem se mostra mais adequada para fins de modelagem do que ocorre em cenas naturais, onde as alterações que acontecem ao longo de um período de tempo tendem a apresentar uma variação contínua em lugar de discreta. Em uma segunda etapa, os graus de pertinência estimados recebem um ajustamento adicional por meio da introdução de informações de contexto espacial. A metodologia proposta foi testada por meio de três experimentos, um empregando uma imagem sintética e dois utilizando imagens reais. A partir da análise quantitativa dos resultados e comparação com estudos semelhantes, comprova-se a adequação da metodologia proposta. / In this dissertation it is proposed a new methodology to land use change detection in remote sensing multitemporal image data. Rather than applying a rigid labeling of the pixels in the image data into two classes (change, no-change), we propose estimating the degrees of membership to classes change and no-change in a fuzzy-like fashion. To this end, a sub-pixel approach is implemented to detect the degree of change in every pixel. This methodology aims at modeling natural scenes in a more realistic way, since changes in natural scenes tend to occur in a continuum rather than in a sharp distinctive way. In a second step, the estimated values for the degrees of membership are further refined by means of spatial context information. Three experiments were performed to test the proposed methodology, one employing synthetic data and two using real image data. From the quantitative analysis of the results and from similar studies we can prove the adequacy of the proposed methodology.

A detecção de mudanças na superfície terrestre é o principal objetivo em aplicações de sensoriamento remoto multitemporal. Sabe-se que imagens adquiridas em datas distintas tendem a ser altamente influenciadas por problemas radiométricos e de registro. Utilizando imagens de fração, obtidas a partir do modelo linear de mistura espectral (MLME), problemas radiométricos podem ser minimizados e a interpretação dos tipos de mudança na superfície terrestre é facilitada, pois as frações têm um significado físico direto. Além disso, interpretações ao nível de subpixel são possíveis. Esta tese propõe três algoritmos – rígido, suave e fuzzy – para a detecção de mudanças entre um par de imagens de fração, gerando mapas de mudança como produtos finais. As propostas requerem a suposição de normalidade multivariada para as diferenças de fração e necessitam de pouca intervenção por parte do analista. A proposta rígida cria mapas de mudança binários seguindo a mesma metodologia de um teste de hipóteses, baseando-se no fato de que os contornos de densidade constante na distribuição normal multivariada são definidos por valores da distribuição qui-quadrado, de acordo com a escolha do nível de confiança. O classificador suave permite gerar estimativas da probabilidade do pixel pertencer à classe de mudança, a partir de um modelo de regressão logística. Essas probabilidades são usadas para criar um mapa de probabilidades de mudança. A abordagem fuzzy é aquela que melhor se adapta ao conceito de pixel mistura, visto que as mudanças no uso e cobertura do solo podem ocorrer em nível de subpixel. Com base nisso, mapas dos graus de pertinência à classe de mudança foram criados. Outras ferramentas matemáticas e estatísticas foram utilizadas, tais como operações morfológicas, curvas ROC e algoritmos de clustering. As três propostas foram testadas utilizando-se imagens sintéticas e reais (Landsat-TM) e avaliadas qualitativa e quantitativamente. Os resultados indicam a viabilidade da utilização de imagens de fração em estudos de detecção de mudanças por meio dos algoritmos propostos. / Land cover change detection is a major goal in multitemporal remote sensing applications. It is well known that images acquired on different dates tend to be highly influenced by radiometric differences and registration problems. Using fraction images, obtained from the linear model of spectral mixing (LMSM), radiometric problems can be minimized and the interpretation of changes in land cover is facilitated because the fractions have a physical meaning. Furthermore, interpretations at the subpixel level are possible. This thesis presents three algorithms – hard, soft and fuzzy – for detecting changes between a pair of fraction images. The algorithms require multivariate normality for the differences among fractions and very little intervention by the analyst. The hard algorithm creates binary change maps following the same methodology of hypothesis testing, based on the fact that the contours of constant density are defined by chi-square values, according to the choice of the probability level. The soft one allows for the generation of estimates of the probability of each pixel belonging to the change class by using a logistic regression model. These probabilities are used to create a map of change probabilities. The fuzzy approach is the one that best fits the concept behind the fraction images because the changes in land cover can occurr at a subpixel level. Based on these algorithms, maps of membership degrees were created. Other mathematical and statistical techniques were also used, such as morphological operations, ROC curves and a clustering algorithm. The algorithms were tested using synthetic and real images (Landsat-TM) and the results were analyzed qualitatively and quantitatively. The results indicate that fraction images can be used in change detection studies by using the proposed algorithms.

Pour l’analyse de la dynamique territoriale il est fondamental une grande quantité de données et leur intégration avec des données spatiales et statistiques facilite ce processus. La thèse se propose d’analyser la dynamique de la superficie agricole des municipalités du Territoire de la Citoyenneté Região Central-RS (TCRCRS) qui fait partie d’une politique publique spatiale de territoires citoyens. Cette dynamique sera analysée à partir d’une technique de détection de changement connue par Rotation Contrôlée par Axe de Non Changement - RCEN. Ainsi, la thèse a comme objectif évaluer les altérations subies par la superficie agricole au long de la période 1985 / 2010 dans les municipalités du TCRCRS en utilisant l’algorithme RCEN. Les étapes méthodologiques suivantes ont été implémentées: utilisation d’images différentes pour l’obtention de pixeis échantillons de non changement; analyse qualitative de l’organisation de la superficie agricole pour les municipalités de Cacequi, Santiago et Tupanciretã, sélectionnés en raison de leur localisation parmi des différentes sous-unités de paysage dans l’État; définition des seuils pour la délimitation des classes thématiques de l’organisation spatiale de la superficie agricole et évaluation de la fiabilité des résultats de la technique RCEN, utilisée pour déterminer la précision de la classification supervisée des images TM Landsat 5 par une matrice de concaténation. La matrice est basée sur l’algèbre de cartes de façon à obtenir une image numérique finale qui exprime toutes les possibilités de l’espace échantillon. Les résultats ont montré que, avec 1% de signification, la technique RCEN peut être utilisée pour détecter la dynamique dans la superficie agricole en utilisant les seuils de vigueur végétatif de l’IDETEC comparés aux résultats des répertoires Normalized Difference Vegetation Index (NDVI), qui a été obtenu tout en considérant le total de pluies antérieures au passage du senseur, variable qui interfère sur les valeurs moyennes de NDVI. Des images de détections de changements (IDETEC) ont été engendrées pour analyser les cultures agricoles d’hiver et d’été, s’obtenant 99% de confiance en les images choisies pour la distribution spatiale des classes définies par l’adoption des seuils de ( - 0,5σ ; – 1,5σ ; + 0,5σ ; + 1,5σ), en prenant comme point central de la classe de non changement. Les images de détection de changements ont permis d’estimer et de comparer les classes de l’IDETEC avec les estimations du total d’aire plantée de cultures temporaires et des cultures agricoles de riz, avoine, maïs, soja et blé. Les aires obtenues par l’IDETEC à Tupanciretã ont surestimé l’aire agricole présentée par l’IBGE dans les images d’été avec des variations en pourcentages entre 1,11% dans l’IDETEC 1994/2009 et 8,13% dans l’IDETEC 2004/2010 ; pour les images d’hiver l’altération a été de 9,46% dans l’IDETEC 1989/2007 et de 3,44% dans l’IDETEC 1996/2005. À la municipalité de Cacequi, les variations en pourcentages de cultures temporaires ont été surestimées dans les images d’été en 7,71% dans l’IDETEC 1986/2006 et 20,47% dans l’IDETEC 1993/2005 et sous-estimées dans les images d’hiver en 9,42% dans l’IDETEC 1985/2003 et en 18,11% dans l’IDETEC 1996/2007. À Santiago elles ont été sous-estimées pour la période d’été en 24,76% dans l’IDETEC 1984/2009, pour la période d’hiver en 10,52% dans l’IDETEC 1996/2005 et surestimées en 8,23% dans l’IDETEC 2004/2010 et en 26,12% pour l’image d’hiver IDETEC 1989/2007. La technique RCEN a prouvé être capable d’évaluer des altérations dans la superficie agricole de cultures annuelles pour les municipalités de Cacequi, Santiago et Tupanciretã. / Para análise da dinâmica territorial, é fundamental uma grande quantidade de dados e a integração com dados espaciais e estatísticos, facilita esse processo. A tese propõe analisar, a dinâmica da superfície agrícola de municípios do Território da Cidadania Região Central-RS (TCRCRS), território este que, faz parte de uma política pública de territórios da cidadania. Essa dinâmica, será analisada a partir de uma técnica de detecção de mudança, conhecida por Rotação Controlada por Eixo de Não Mudança - RCEN. Assim, a tese objetiva avaliar as alterações na superfície agrícola, no período de 1985 a 2010, em municípios do TCRCRS, utilizando o algoritmo RCEN. As seguintes etapas metodológicas, foram implementadas: utilização de imagens diferentes, para obtenção de pixeis amostrais de não mudança; análise qualitativamente da organização da superfície agrícola, para os municípios de Cacequi, Santiago e Tupanciretã, selecionados por sua localização em diferentes subunidades de paisagem no Estado; definição dos limiares, para delimitação das classes temáticas da organização espacial da superfície agrícola e, avaliação da confiabilidade dos resultados da técnica RCEN, utilizada pra determinar, a precisão da classificação supervisionada das imagens TM Landsat 5, através de uma matriz de concatenação. A matriz, é baseada em álgebra de mapas de tal maneira, a obter uma imagem numérica final que, expresse todas as possibilidades do espaço amostral. Os resultados, mostraram que, com 1% de significância, a técnica RCEN pode ser utilizada, para detectar a dinâmica na superfície agrícola, utilizando limiares de vigor vegetativo da IDETEC, comparados aos resultados dos índices Normalized Difference Vegetation Index (NDVI), que foi obtido considerando, o total de chuvas antecedentes a passagem do sensor, que é uma variável que interfere, nos valores médios de NDVI. Foram geradas imagens de detecção de mudanças (IDETEC), para analisar culturas agrícolas de inverno e de verão, obtendo-se 99% de confiança nas imagens selecionadas, para a distribuição espacial das classes definidas pela adoção dos limiares de ( - 0,5σ; – 1,5σ ; + 0,5σ; + 1,5σ), utilizando a como ponto central da classe de não mudança. As imagens de detecção de mudanças, permitiram estimar e comparar as classes da IDETEC, com as estimativas do total de área plantada, de lavouras temporárias, das culturas agrícolas de arroz, aveia, milho, soja e trigo As áreas obtidas pela IDETEC em Tupanciretã, superestimaram a área agrícola, apresentada pelo IBGE, nas imagens de verão com variações percentuais entre 1,11% na IDETEC 1994/2009 e 8,13% na IDETEC 2004/2010, para as imagens de inverno, a alteração foi de 9,46% na IDETEC 1989/2007 e de 3,44% na IDETEC 1996/2005. No município de Cacequi, as variações percentuais de lavouras temporárias foram superestimadas nas imagens de verão em 7,71% na IDETEC 1986/2006 e 20,47% na IDETEC 1993/2005 e, subestimadas nas imagens de inverno em 9,42% na IDETEC 1985/2003 e em 18,11% na IDETEC 1996/2007. Em Santiago, foram subestimadas para o período de verão em 24,76% na IDETEC 1984/2009 e, para o período de inverno em 10,52%, na IDETEC 1996/2005 e superestimadas em 8,23% na IDETEC 2004/2010 e, em 26,12% para a imagem de inverno IDETEC 1989/2007. A técnica RCEN, demonstrou ser capaz de estimar alterações na superfície agrícola, de culturas anuais para os municípios de Cacequi, Santiago e Tupanciretã. / For the analysis of territorial dynamics, a great amount of data is fundamental, and the integration of spatial and statistical data facilitates this process. This thesis proposes to analyze the dynamic of the agricultural surface in the Citizenship Country in the Central Area of Rio Grande do Sul (CCCARS), a country that is part of a public policy of citizenship countries. This dynamics will be analyzed by a change detection technique, known as Rotation Controlled of Non-change Axis (RCNA).Thus, this thesis aims to evaluate the alterations in the agricultural surface, in the period from 1985 to 2010, in CCCARS cities, using the RCNA algorithm. The following methodological steps were implemented: the use of different images in order to obtain non-change sampling pixeis; qualitative analysis of the organization of the agricultural surface in the cities of Cacequi, Santiago and Tupanciretã, which were selected due to their location in different subunits of landscapes in the State; determination of thresholds for the delimitation of thematic clusters in the spatial organization of the agricultural surface; and the evaluation of the reliability of the results of RCNA technique, which was used to determine the accuracy of the supervised classification of Landsat TM 5 images through a concatenating matrix. The matrix is based on the map algebra in such manner that expresses all the possibilities of the sampling space. The results showed that with 1% of significance, the RCNA technique can be used to detect the dynamics of the agricultural surface using threshold of the vigor of the vegetative growth compared with the results of Normalized Difference Vegetation Index (NDVI), which were obtained considering the total amount of rain previous to the sensor scanning, which is a variable that interferes in the medium values of NDVI. It was created images of changes detection (IDETEC) in order to analyze summer and winter agricultural crops, obtaining 99% of reliability on the selected images, for the special distribution of the defined clusters by the adoption of the threshold values of de ( - 0,5σ; – 1,5σ ; + 0,5σ; + 1,5σ), using the as a central point of nonchange cluster. The change detection images enabled to estimate and compare IDETEC clusters with the estimate of the total planted area of temporary farm, agricultural crop of rice, oat, corn, soybean and wheat. The areas obtained by IDETEC in Tupanciretã overestimated the agricultural area presented by IBGE, with summer images with percentage variance among 1,11% on IDETEC 1994/2009 and 8,13% on IDETEC 2004/2010, for the winter images, the alteration was of 9,46% on IDETEC 1989/2007 and of 3,44% on IDETEC 1996/2007. In the city of Cacequi, the percentage variance of the temporary farms were overestimated on the summer images in 7,71% on IDETEC 1986/2006 and 20,47% on IDETEC 1993/2005 and , and overestimated on the winter images in 9,42% on IDETEC 1985/2003 and in 18,11% on IDETEC 1996/2007. In Santiago, they were underestimated for the summer period in 24,76% on IDETEC 1984/2009 and , for the winter period in 10,52%, on IDETEC 1996/2005 e and overestimated in 8,23% on IDETEC 2004/2010 and, in 26,12% for the winter image IDETEC 1989/2007. The RCNA technique showed itself to be capable of estimating the agricultural surface alteration in annual crops in the cities of Cacequi, Santiago e Tupanciretã.

Les travaux de cette thèse concernent la détection des changements dans des séries chronologiques de photographies de paysages prises depuis le sol. Ce contexte de comparaison d'images successives est celui que rencontrent les géographes de l'environnement qui ont recours aux observatoires photographiques du paysage. Ces outils d'analyse et d'aide à la décision sont des bases de données de photographies constituées selon une méthodologie stricte de rephotographie de la même scène, à des pas de temps réguliers. Le nombre de clichés est parfois très important, et l'analyse humaine fastidieuse et relativement imprécise, aussi un outil automatisant la comparaison de photos de paysage deux à deux pour mettre en évidence les changements serait une aide considérable dans l'exploitation des observatoires photographiques du paysage. Bien entendu, les variations dans l'éclairement, la saisonnalité, l'heure du jour, produisent fatalement des clichés entièrement différents à l'échelle du pixel. Notre objectif était donc de concevoir un système robuste face à ces changements mineurs, mais capable de détecter les changements pertinents de l'environnement. De nombreux travaux autour de la détection des changements ont été effectués pour des images provenant de satellites. Mais l'utilisation d'appareils photographiques numériques classiques depuis le sol pose des problèmes spécifiques comme la limitation du nombre de bandes spectrales et la forte variation de profondeur dans une même image qui induit des apparences différentes des mêmes catégories d'objets en fonction de leurs positions dans la scène. Dans un premier temps, nous avons exploré la voie de la détection automatique des changements. Nous avons proposé une méthode reposant sur le recalage et la sur-segmentation des images en superpixels. Ces derniers sont ensuite décrits par leur niveau de gris moyen ainsi que par leur texture au travers d'une représentation sous la forme d'histogrammes de textons. La distance de Mahalanobis entre ces descripteurs permet de comparer les superpixels correspondants entre deux images prises à des dates différentes. Nous avons évalué les performances de cette approche sur des images de l'observatoire photographique du paysage constitué lors de la construction de l'autoroute A89. Parmi les méthodes de segmentation utilisées pour produire les superpixels, les expérimentations que nous avons menées ont mis en évidence le bon comportement de la méthode de segmentation d'Achanta. La pertinence d'un changement étant fortement liée à l'application visée, nous avons exploré dans un second temps une piste faisant intervenir l'utilisateur. Nous avons proposé une méthode interactive de détection des changements reposant sur une phase d'apprentissage. Afin de détecter les changements entre deux images, l'utilisateur désigne, grâce à un outil de sélection, des échantillons constitués d'ensembles de pixels correspondant à des zones de changement et à des zones d'absence de changement. Chaque couple de pixels correspondants, c'est-à-dire situés au même endroit dans les deux images, est décrit par un vecteur de 16 valeurs principalement calculées à partir de l'image des dissemblances. Cette dernière est obtenue en mesurant, pour chaque couple de pixels correspondants, la dissemblance des niveaux de gris de leurs voisinages. Les échantillons désignés par l'utilisateur permettent de constituer des données d'apprentissage qui sont utilisées pour entraîner un classifieur. Parmi les méthodes de classification évaluées, les résultats expérimentaux montrent que les forêts d'arbres décisionnels donnent les meilleurs résultats sur les séries photographiques que nous avons utilisées. / This work deals with change detection from chronological series of photographs acquired from the ground. This context of consecutive images comparison is the one encountered in the field of integrated geography where photographic landscape observatories are widely used. These tools for analysis and decision-making consist of databases of photographic images obtained by strictly rephotographing the same scene at regular time intervals. With a large number of images, the human analysis is tedious and inaccurate. So a tool for automatically comparing pairs of landscape photographs in order to highlight changes would be a great help for exploiting photographic landscape observatories. Obviously, lighting variations, seasonality, time of day induce completely different images at the pixel level. Our goal is to design a system which would be robust to these insignificant changes and able to detect relevant changes of the scene. Numerous studies have been conducted on change detection from satellite images. But the utilization of classic digital cameras from the ground raise some specific problems like the limitation of the spectral band number and the strong variation of the depth in a same image which induces various appearance of the same object categories depending on their position in the scene. In the first part of our work, we investigate the track of automatic change detection. We propose a method lying on the registration and the over-segmentation of the images into superpixels. Then we describe each superpixel by its texture using texton histogram and its gray-level mean. A distance measure, such as Mahalanobis distance, allows to compare corresponding superpixels between two images acquired at different dates. We evaluate the performance of the proposed approach on images taken from the photographic landscape observatory produced during the construction of the French A89 highway. Among the image segmentation methods we have tested for superpixel extraction, our experiments show the relatively good behavior of Achanta segmentation method. The relevance of a change is strongly related to the intended application, we thus investigate a second track involving a user intervention. We propose an interactive change detection method based on a learning step. In order to detect changes between two images, the user designates with a selection tool some samples consisting of pixel sets in "changed" and "unchanged" areas. Each corresponding pixel pair, i.e., located at the same coordinates in the two images, is described by a 16-dimensional feature vector mainly calculated from the dissimilarity image. The latter is computed by measuring, for each corresponding pixel pair, the dissimilarity of the gray-levels of the neighbors of the two pixels. Samples selected by the user are used as learning data to train a classifier. Among the classification methods we have tried, experimental results indicate that random forests give the better results for the tested image series.

Context: We identify a material or thing that can be seen and touched in the world as having structures at both coarser and finer levels of scale. Scaling problem presents in a branch of science concerned with the description, prediction understanding of natural phenomena and visual arts. A moon, for instance, may appear as having a roughly round shape is much larger than stars when seen from the earth. In the closer look, the moon is much smaller than the stars. The fact that objects in the world appear in different ways depending upon the scale of observation has important implications when analyzing measured data, such as images, with automatic methods [1]. The type of information we are seeking from a one-dimensional signal or two-dimensional image is only possible when we have the right amount of scale for the structure of an image or signal data. In many modern applications, the right scale need not be obvious at all, and we all need a complete mathematical analysis on this scaling problem. This thesis is shown how a mathematical theory is formulated when data or signal is describing at different scales. Objectives: The subtle patterns deforming in data that can foretell of a scaling problem? The main objectives of this thesis are to address the dynamic scaling pattern problem in computers and study the different methods, described in the latest issue of Science, are designed to identify the patterns in data. Method: The research methodology used in this thesis is the Fractional Fourier Transform. To recognize the pattern for a different level of scale to one or many components, we take the position and size of the object and perform the transform operation in any transform angle and deform the component by changing to another angle which influences the frequency, phase, and magnitude.  Results: We show that manipulation of Fractional Fourier transform can be used as a pattern recognition system. The introduced model has the flexibility to encode patterns to both time and frequency domain. We present a detailed structure of a dynamic pattern scaling problem. Furthermore, we show successful recognition results even though one or many components deformed to different levels using one-dimensional and two-dimensional patterns. Conclusions: The proposed algorithm FrFT has shown some advantages over traditional FFT due to its competitive performance in studying the pattern changes. This research work investigated that simulating the dynamic pattern scaling problem using FrFT. The Fractional Fourier transform does not do the scaling. Manipulating the Fractional Fourier transform can be helpful in perceiving the pattern changes. We cannot control the deformation but changing the parameters allow us to see what is happening in time and frequency domain.

Nesta dissertação está proposta uma metodologia para fins de detecção de mudanças do uso do solo em imagens multitemporais de sensoriamento remoto. Em lugar de classificar os pixels de imagens que cobrem uma cena, em duas classes exaustivas e mutuamente excludentes (mudança, não-mudança), propõe-se adotar uma abordagem do tipo fuzzy, na qual são estimados os graus de pertinência às classes mudança e não-mudança. Com este objetivo adota-se aqui uma abordagem em nível de sub-pixel na estimação dos graus de pertinência para cada pixel. Esta abordagem se mostra mais adequada para fins de modelagem do que ocorre em cenas naturais, onde as alterações que acontecem ao longo de um período de tempo tendem a apresentar uma variação contínua em lugar de discreta. Em uma segunda etapa, os graus de pertinência estimados recebem um ajustamento adicional por meio da introdução de informações de contexto espacial. A metodologia proposta foi testada por meio de três experimentos, um empregando uma imagem sintética e dois utilizando imagens reais. A partir da análise quantitativa dos resultados e comparação com estudos semelhantes, comprova-se a adequação da metodologia proposta. / In this dissertation it is proposed a new methodology to land use change detection in remote sensing multitemporal image data. Rather than applying a rigid labeling of the pixels in the image data into two classes (change, no-change), we propose estimating the degrees of membership to classes change and no-change in a fuzzy-like fashion. To this end, a sub-pixel approach is implemented to detect the degree of change in every pixel. This methodology aims at modeling natural scenes in a more realistic way, since changes in natural scenes tend to occur in a continuum rather than in a sharp distinctive way. In a second step, the estimated values for the degrees of membership are further refined by means of spatial context information. Three experiments were performed to test the proposed methodology, one employing synthetic data and two using real image data. From the quantitative analysis of the results and from similar studies we can prove the adequacy of the proposed methodology.

Pour l’analyse de la dynamique territoriale il est fondamental une grande quantité de données et leur intégration avec des données spatiales et statistiques facilite ce processus. La thèse se propose d’analyser la dynamique de la superficie agricole des municipalités du Territoire de la Citoyenneté Região Central-RS (TCRCRS) qui fait partie d’une politique publique spatiale de territoires citoyens. Cette dynamique sera analysée à partir d’une technique de détection de changement connue par Rotation Contrôlée par Axe de Non Changement - RCEN. Ainsi, la thèse a comme objectif évaluer les altérations subies par la superficie agricole au long de la période 1985 / 2010 dans les municipalités du TCRCRS en utilisant l’algorithme RCEN. Les étapes méthodologiques suivantes ont été implémentées: utilisation d’images différentes pour l’obtention de pixeis échantillons de non changement; analyse qualitative de l’organisation de la superficie agricole pour les municipalités de Cacequi, Santiago et Tupanciretã, sélectionnés en raison de leur localisation parmi des différentes sous-unités de paysage dans l’État; définition des seuils pour la délimitation des classes thématiques de l’organisation spatiale de la superficie agricole et évaluation de la fiabilité des résultats de la technique RCEN, utilisée pour déterminer la précision de la classification supervisée des images TM Landsat 5 par une matrice de concaténation. La matrice est basée sur l’algèbre de cartes de façon à obtenir une image numérique finale qui exprime toutes les possibilités de l’espace échantillon. Les résultats ont montré que, avec 1% de signification, la technique RCEN peut être utilisée pour détecter la dynamique dans la superficie agricole en utilisant les seuils de vigueur végétatif de l’IDETEC comparés aux résultats des répertoires Normalized Difference Vegetation Index (NDVI), qui a été obtenu tout en considérant le total de pluies antérieures au passage du senseur, variable qui interfère sur les valeurs moyennes de NDVI. Des images de détections de changements (IDETEC) ont été engendrées pour analyser les cultures agricoles d’hiver et d’été, s’obtenant 99% de confiance en les images choisies pour la distribution spatiale des classes définies par l’adoption des seuils de ( - 0,5σ ; – 1,5σ ; + 0,5σ ; + 1,5σ), en prenant comme point central de la classe de non changement. Les images de détection de changements ont permis d’estimer et de comparer les classes de l’IDETEC avec les estimations du total d’aire plantée de cultures temporaires et des cultures agricoles de riz, avoine, maïs, soja et blé. Les aires obtenues par l’IDETEC à Tupanciretã ont surestimé l’aire agricole présentée par l’IBGE dans les images d’été avec des variations en pourcentages entre 1,11% dans l’IDETEC 1994/2009 et 8,13% dans l’IDETEC 2004/2010 ; pour les images d’hiver l’altération a été de 9,46% dans l’IDETEC 1989/2007 et de 3,44% dans l’IDETEC 1996/2005. À la municipalité de Cacequi, les variations en pourcentages de cultures temporaires ont été surestimées dans les images d’été en 7,71% dans l’IDETEC 1986/2006 et 20,47% dans l’IDETEC 1993/2005 et sous-estimées dans les images d’hiver en 9,42% dans l’IDETEC 1985/2003 et en 18,11% dans l’IDETEC 1996/2007. À Santiago elles ont été sous-estimées pour la période d’été en 24,76% dans l’IDETEC 1984/2009, pour la période d’hiver en 10,52% dans l’IDETEC 1996/2005 et surestimées en 8,23% dans l’IDETEC 2004/2010 et en 26,12% pour l’image d’hiver IDETEC 1989/2007. La technique RCEN a prouvé être capable d’évaluer des altérations dans la superficie agricole de cultures annuelles pour les municipalités de Cacequi, Santiago et Tupanciretã. / Para análise da dinâmica territorial, é fundamental uma grande quantidade de dados e a integração com dados espaciais e estatísticos, facilita esse processo. A tese propõe analisar, a dinâmica da superfície agrícola de municípios do Território da Cidadania Região Central-RS (TCRCRS), território este que, faz parte de uma política pública de territórios da cidadania. Essa dinâmica, será analisada a partir de uma técnica de detecção de mudança, conhecida por Rotação Controlada por Eixo de Não Mudança - RCEN. Assim, a tese objetiva avaliar as alterações na superfície agrícola, no período de 1985 a 2010, em municípios do TCRCRS, utilizando o algoritmo RCEN. As seguintes etapas metodológicas, foram implementadas: utilização de imagens diferentes, para obtenção de pixeis amostrais de não mudança; análise qualitativamente da organização da superfície agrícola, para os municípios de Cacequi, Santiago e Tupanciretã, selecionados por sua localização em diferentes subunidades de paisagem no Estado; definição dos limiares, para delimitação das classes temáticas da organização espacial da superfície agrícola e, avaliação da confiabilidade dos resultados da técnica RCEN, utilizada pra determinar, a precisão da classificação supervisionada das imagens TM Landsat 5, através de uma matriz de concatenação. A matriz, é baseada em álgebra de mapas de tal maneira, a obter uma imagem numérica final que, expresse todas as possibilidades do espaço amostral. Os resultados, mostraram que, com 1% de significância, a técnica RCEN pode ser utilizada, para detectar a dinâmica na superfície agrícola, utilizando limiares de vigor vegetativo da IDETEC, comparados aos resultados dos índices Normalized Difference Vegetation Index (NDVI), que foi obtido considerando, o total de chuvas antecedentes a passagem do sensor, que é uma variável que interfere, nos valores médios de NDVI. Foram geradas imagens de detecção de mudanças (IDETEC), para analisar culturas agrícolas de inverno e de verão, obtendo-se 99% de confiança nas imagens selecionadas, para a distribuição espacial das classes definidas pela adoção dos limiares de ( - 0,5σ; – 1,5σ ; + 0,5σ; + 1,5σ), utilizando a como ponto central da classe de não mudança. As imagens de detecção de mudanças, permitiram estimar e comparar as classes da IDETEC, com as estimativas do total de área plantada, de lavouras temporárias, das culturas agrícolas de arroz, aveia, milho, soja e trigo As áreas obtidas pela IDETEC em Tupanciretã, superestimaram a área agrícola, apresentada pelo IBGE, nas imagens de verão com variações percentuais entre 1,11% na IDETEC 1994/2009 e 8,13% na IDETEC 2004/2010, para as imagens de inverno, a alteração foi de 9,46% na IDETEC 1989/2007 e de 3,44% na IDETEC 1996/2005. No município de Cacequi, as variações percentuais de lavouras temporárias foram superestimadas nas imagens de verão em 7,71% na IDETEC 1986/2006 e 20,47% na IDETEC 1993/2005 e, subestimadas nas imagens de inverno em 9,42% na IDETEC 1985/2003 e em 18,11% na IDETEC 1996/2007. Em Santiago, foram subestimadas para o período de verão em 24,76% na IDETEC 1984/2009 e, para o período de inverno em 10,52%, na IDETEC 1996/2005 e superestimadas em 8,23% na IDETEC 2004/2010 e, em 26,12% para a imagem de inverno IDETEC 1989/2007. A técnica RCEN, demonstrou ser capaz de estimar alterações na superfície agrícola, de culturas anuais para os municípios de Cacequi, Santiago e Tupanciretã. / For the analysis of territorial dynamics, a great amount of data is fundamental, and the integration of spatial and statistical data facilitates this process. This thesis proposes to analyze the dynamic of the agricultural surface in the Citizenship Country in the Central Area of Rio Grande do Sul (CCCARS), a country that is part of a public policy of citizenship countries. This dynamics will be analyzed by a change detection technique, known as Rotation Controlled of Non-change Axis (RCNA).Thus, this thesis aims to evaluate the alterations in the agricultural surface, in the period from 1985 to 2010, in CCCARS cities, using the RCNA algorithm. The following methodological steps were implemented: the use of different images in order to obtain non-change sampling pixeis; qualitative analysis of the organization of the agricultural surface in the cities of Cacequi, Santiago and Tupanciretã, which were selected due to their location in different subunits of landscapes in the State; determination of thresholds for the delimitation of thematic clusters in the spatial organization of the agricultural surface; and the evaluation of the reliability of the results of RCNA technique, which was used to determine the accuracy of the supervised classification of Landsat TM 5 images through a concatenating matrix. The matrix is based on the map algebra in such manner that expresses all the possibilities of the sampling space. The results showed that with 1% of significance, the RCNA technique can be used to detect the dynamics of the agricultural surface using threshold of the vigor of the vegetative growth compared with the results of Normalized Difference Vegetation Index (NDVI), which were obtained considering the total amount of rain previous to the sensor scanning, which is a variable that interferes in the medium values of NDVI. It was created images of changes detection (IDETEC) in order to analyze summer and winter agricultural crops, obtaining 99% of reliability on the selected images, for the special distribution of the defined clusters by the adoption of the threshold values of de ( - 0,5σ; – 1,5σ ; + 0,5σ; + 1,5σ), using the as a central point of nonchange cluster. The change detection images enabled to estimate and compare IDETEC clusters with the estimate of the total planted area of temporary farm, agricultural crop of rice, oat, corn, soybean and wheat. The areas obtained by IDETEC in Tupanciretã overestimated the agricultural area presented by IBGE, with summer images with percentage variance among 1,11% on IDETEC 1994/2009 and 8,13% on IDETEC 2004/2010, for the winter images, the alteration was of 9,46% on IDETEC 1989/2007 and of 3,44% on IDETEC 1996/2007. In the city of Cacequi, the percentage variance of the temporary farms were overestimated on the summer images in 7,71% on IDETEC 1986/2006 and 20,47% on IDETEC 1993/2005 and , and overestimated on the winter images in 9,42% on IDETEC 1985/2003 and in 18,11% on IDETEC 1996/2007. In Santiago, they were underestimated for the summer period in 24,76% on IDETEC 1984/2009 and , for the winter period in 10,52%, on IDETEC 1996/2005 e and overestimated in 8,23% on IDETEC 2004/2010 and, in 26,12% for the winter image IDETEC 1989/2007. The RCNA technique showed itself to be capable of estimating the agricultural surface alteration in annual crops in the cities of Cacequi, Santiago e Tupanciretã.

A detecção de mudanças na superfície terrestre é o principal objetivo em aplicações de sensoriamento remoto multitemporal. Sabe-se que imagens adquiridas em datas distintas tendem a ser altamente influenciadas por problemas radiométricos e de registro. Utilizando imagens de fração, obtidas a partir do modelo linear de mistura espectral (MLME), problemas radiométricos podem ser minimizados e a interpretação dos tipos de mudança na superfície terrestre é facilitada, pois as frações têm um significado físico direto. Além disso, interpretações ao nível de subpixel são possíveis. Esta tese propõe três algoritmos – rígido, suave e fuzzy – para a detecção de mudanças entre um par de imagens de fração, gerando mapas de mudança como produtos finais. As propostas requerem a suposição de normalidade multivariada para as diferenças de fração e necessitam de pouca intervenção por parte do analista. A proposta rígida cria mapas de mudança binários seguindo a mesma metodologia de um teste de hipóteses, baseando-se no fato de que os contornos de densidade constante na distribuição normal multivariada são definidos por valores da distribuição qui-quadrado, de acordo com a escolha do nível de confiança. O classificador suave permite gerar estimativas da probabilidade do pixel pertencer à classe de mudança, a partir de um modelo de regressão logística. Essas probabilidades são usadas para criar um mapa de probabilidades de mudança. A abordagem fuzzy é aquela que melhor se adapta ao conceito de pixel mistura, visto que as mudanças no uso e cobertura do solo podem ocorrer em nível de subpixel. Com base nisso, mapas dos graus de pertinência à classe de mudança foram criados. Outras ferramentas matemáticas e estatísticas foram utilizadas, tais como operações morfológicas, curvas ROC e algoritmos de clustering. As três propostas foram testadas utilizando-se imagens sintéticas e reais (Landsat-TM) e avaliadas qualitativa e quantitativamente. Os resultados indicam a viabilidade da utilização de imagens de fração em estudos de detecção de mudanças por meio dos algoritmos propostos. / Land cover change detection is a major goal in multitemporal remote sensing applications. It is well known that images acquired on different dates tend to be highly influenced by radiometric differences and registration problems. Using fraction images, obtained from the linear model of spectral mixing (LMSM), radiometric problems can be minimized and the interpretation of changes in land cover is facilitated because the fractions have a physical meaning. Furthermore, interpretations at the subpixel level are possible. This thesis presents three algorithms – hard, soft and fuzzy – for detecting changes between a pair of fraction images. The algorithms require multivariate normality for the differences among fractions and very little intervention by the analyst. The hard algorithm creates binary change maps following the same methodology of hypothesis testing, based on the fact that the contours of constant density are defined by chi-square values, according to the choice of the probability level. The soft one allows for the generation of estimates of the probability of each pixel belonging to the change class by using a logistic regression model. These probabilities are used to create a map of change probabilities. The fuzzy approach is the one that best fits the concept behind the fraction images because the changes in land cover can occurr at a subpixel level. Based on these algorithms, maps of membership degrees were created. Other mathematical and statistical techniques were also used, such as morphological operations, ROC curves and a clustering algorithm. The algorithms were tested using synthetic and real images (Landsat-TM) and the results were analyzed qualitatively and quantitatively. The results indicate that fraction images can be used in change detection studies by using the proposed algorithms.