No description available.

Within the field of pattern recognition (PR) a very active area is the clustering and classification of multispectral data, which basically aims to allocate the right class of ground category to a reflectance or radiance signal. Generally, the problem complexity is related to the incorporation of spatial characteristics that are complementary to the nonlinearities of land surface process heterogeneity, remote sensing effects and multispectral features. The present research describes the application of learning machine methods to accomplish the above task by inducting a relationship between the spectral response of farms’ land cover, and their farming system typology from a representative set of instances. Such methodologies are not traditionally used in crop-livestock studies. Nevertheless, this study shows that its application leads to simple and theoretically robust classification models. The study has covered the following phases: a)geovisualization of crop-livestock systems; b)feature extraction of both multispectral and attributive data and; c)supervised farm classification. The first is a complementary methodology to represent the spatial feature intensity of farming systems in the geographical space. The second belongs to the unsupervised learning field, which mainly involves the appropriate description of input data in a lower dimensional space. The last is a method based on statistical learning theory, which has been successfully applied to supervised classification problems and to generate models described by implicit functions. In this research the performance of various kernel methods applied to the representation and classification of crop-livestock systems described by multispectral response is studied and compared. The data from those systems include linear and nonlinearly separable groups that were labelled using multidimensional attributive data. Geovisualization findings show the existence of two well-defined farm populations within the whole study area; and three subgroups in relation to the Guarico section. The existence of these groups was confirmed by both hierarchical and kernel clustering methods, and crop-livestock systems instances were segmented and labeled into farm typologies based on: a)milk and meat production; b)reproductive management; c)stocking rate; and d)crop-forage-forest land use. The minimum set of labeled examples to properly train the kernel machine was 20 instances. Models inducted by training data sets using kernel machines were in general terms better than those from hierarchical clustering methodologies. However, the size of the training data set represents one of the main difficulties to be overcome in permitting the more general application of this technique in farming system studies. These results attain important implications for large scale monitoring of crop-livestock system; particularly to the establishment of balanced policy decision, intervention plans formulation, and a proper description of target typologies to enable investment efforts to be more focused at local issues.

Heterogeneity in forest vegetation monitoring with remote sensing Abtract The main aim of this diploma thesis is to examine the suitability of various classification approaches for forest vegetation categorization using Landsat 8 satellite imagery. Two satellite images acquired during vegetative period (8th March, 27th July 2013) were chosen. The overall goal of the study is to explore the potential of using statistical methods to obtain information about forest heterogeneity in a given territory. Chosen study sites are defined by administrative boundaries of selected municipalities from South and Central Bohemia located within following municipalities with extended powers - Blatná, Milevsko, Písek, Příbram and Sedlčany. Supervised and unsupervised classifications were used based on obtained training areas and orthophoto. The definition of chosen classes (coniferous forests, mixed forests, ecotones, structurally homogeneous deciduous forests and structurally heterogeneous deciduous forests) was identical with the categories used in "Project MT 11425-5/2010 The Mapping of Natural Zoonoses Focal Points, Transferable on Humans in the Czech Republic and Their Changes Affected by the Modification of Climate". Due to large amount of training datasets obtained from field survey, ortophoto and spectral analysis,...

The dissertation deals with utilisation of multispectral optical measurement for data fusion that may be used for visual telepresence and indoor/outdoor mapping by heterogeneous mobile robotic system. Optical proximity sensors, thermal imagers, and tricolour cameras are used for the fusion. The described algorithms are optimised to work in real-time and implemented on CASSANDRA robotic system made by our robotic research group.

Submitted by GEORGE DEROCO MARTINS null (deroco87@hotmail.com) on 2017-02-08T14:17:06Z No. of bitstreams: 1 martins_gd_dr_prud.pdf: 4750620 bytes, checksum: 2b77564f0f8c206ce23e6d7ca4bac5d6 (MD5) / Approved for entry into archive by LUIZA DE MENEZES ROMANETTO (luizamenezes@reitoria.unesp.br) on 2017-02-13T18:01:23Z (GMT) No. of bitstreams: 1 martins_gd_dr_prud.pdf: 4750620 bytes, checksum: 2b77564f0f8c206ce23e6d7ca4bac5d6 (MD5) / Made available in DSpace on 2017-02-13T18:01:23Z (GMT). No. of bitstreams: 1 martins_gd_dr_prud.pdf: 4750620 bytes, checksum: 2b77564f0f8c206ce23e6d7ca4bac5d6 (MD5) Previous issue date: 2016-12-19 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) / Os nematoides são importantes fitoparasitas que se constituem em um problema sério para o cultivo do café no Brasil. Como a ocorrência de nematoides no sistema radicular do cafeeiro causa desequilíbrios nutricionais na planta que provocam variações na resposta espectral da folha e define uma configuração espacial característica às áreas infectadas, o objetivo desta pesquisa avaliar o potencial de dados de sensoriamento remoto adquiridos em multiescala para discriminar e mapear o café sadio, em estágio inicial de infecção e severamente infectado. A pesquisa foi desenvolvida em três áreas experimentais, localizadas no sul do estado de Minas Gerais, nas quais foi certificada a ocorrência de nematoides e realizadas medições de variáveis biofísicas e dados hiperespectrais na folha e sobre o dossel da planta. Os dados hiperespectrais também foram utilizados em simulação de bandas dos sensores do RapidEye e OLI/Landsat 8 para identificar as faixas espectrais mais sensíveis para a discriminação de patógenos em plantas de café. Nenhum dos parâmetros biofísicos avaliados discriminou eficientemente as folhas de plantas sadias e infectadas, mas a simulação de bandas indicou que os intervalos espectrais do vermelho, vermelho limítrofe e infravermelho próximos do RapidEye foram complementares para a discriminação de plantas de café sadio e dos dois níveis de infecção. Essas bandas, mais uma imagem NDVI, foram utilizadas na classificação das áreas infectadas por nematoides, a qual definiu a distribuição espacial de café sadio e dos dois níveis de infecção, com uma acurácia global de 78% e coeficiente kappa de 0,71. A classificação não supervisionada da imagem multiespectral OLI/Landsat 8 também definiu as três condições, porém com baixa confiabilidade (coeficiente kappa igual a 0,41). Por outro lado, uma inferência espacial quantitativa da concentração de nematoides/cm³ no solo, a partir de um modelo empírico baseado na imagem RapidEye, apresentou um erro consideravelmente alto (21,89%). / Nematodes are important phytoparasites that constitute a serious issue for coffee cultivation in Brazil. Because root infection by nematodes induces spectral variation in leaves and defines a unique spatial configuration in the cultivation field, the aim of this study is to evaluate the potential of remote sensing data acquired in multiscale to discriminate and map healthy, early infected and severely infected coffee plants. This study was carried out in three experimental areas, located in the in southern Minas Gerais State, in which the occurrence of nematodes was certified and biophysical and hyperspectral measurements of the leaves and on the canopy were made. Hyperspectral data were also used to simulate the bands of the RapidEye and OLI/Landsat 8 sensors to identify the most sensitive spectral ranges for pathogen discrimination in coffee plants. None of the biophysical parameters efficiently discriminated the leaves of healthy and infected plants, but the band simulations indicated that red, red edge and near infrared spectral ranges were complementary to the discrimination of healthy coffee plants and the two levels of infection. These bands, plus an (NDVI) image, were used for a multispectral classification of healthy and nematode-infected areas. The multispectral classification defined the spatial distribution of healthy, early infected and two levels of infection, with an overall accuracy of 78% and kappa coefficient of 0.71. The unsupervised classification of the multispectral image OLI/Landsat 8 also defined the three conditions, but with low reliability (kappa coefficient equal to 0.41). In contrast, a quantitative spatial inference of the soil nematode concentration/cm³, from an empirical model based on the RapidEye image, presented a considerably high error (21.89%).

Les missions spatiales d'observation de la Terre de nouvelle génération telles que Sentinel-2 (préparé par l'Agence Spatiale Européenne ESA dans le cadre du programme Copernicus, auparavant appelé Global Monitoring for Environment and Security ou GMES) ou Venµs, conjointement développé par l'Agence Spatiale Française (Centre National d 'Études Spatiales CNES) et l'Agence Spatiale Israélienne (ISA), vont révolutionner la surveillance de l'environnement d' aujourd'hui avec le rendement de volumes inédits de données en termes de richesse spectrale, de revisite temporelle et de résolution spatiale. Venµs livrera des images dans 12 bandes spectrales de 412 à 910 nm, une répétitivité de 2 jours et une résolution spatiale de 10 m; les satellites jumeaux Sentinel-2 assureront une couverture dans 13 bandes spectrales de 443 à 2200 nm, avec une répétitivité de 5 jours, et des résolutions spatiales de 10 à 60m. La production efficace de cartes d'occupation des sols basée sur l'exploitation de tels volumes d'information pour grandes surfaces est un défi à la fois en termes de coûts de traitement mais aussi de variabilité des données. En général, les méthodes classiques font soit usage des approches surveillées (trop coûteux en termes de travaux manuels pour les grandes surfaces), ou soit ciblent des modèles locaux spécialisés pour des problématiques précises (ne s'appliquent pas à autres terrains ou applications), ou comprennent des modèles physiques complexes avec coûts de traitement rédhibitoires. Ces approches existantes actuelles sont donc inefficaces pour l'exploitation du nouveau type de données que les nouvelles missions fourniront, et un besoin se fait sentir pour la mise en œuvre de méthodes précises, rapides et peu supervisées qui permettent la généralisation à l'échelle de grandes zones avec des résolutions élevées. Afin de permettre l'exploitation des volumes de données précédemment décrits, l'objectif de ce travail est la conception et validation d'une approche entièrement automatique qui permet l'estimation de la couverture terrestre de grandes surfaces avec imagerie d'observation de la Terre de haute résolution spatiale, spectrale et temporelle, généralisable à des paysages différents, et offrant un temps de calcul opérationnel avec ensembles de données satellitaires simulés, en préparation des prochaines missions. Cette approche est basée sur l'intégration d'algorithmes de traitement de données, tels que les techniques d'apprentissage de modèles et de classification, et des connaissances liées à l'occupation des sols sur des questions écologiques et agricoles, telles que les variables avec un impact sur la croissance de la végétation ou les pratiques de production. Par exemple, la nouvelle introduction de température comme axe temporel pour un apprentissage des modèles ultérieurs intègre un facteur établi de la croissance de la végétation à des techniques d'apprentissage automatiques pour la caractérisation des paysages. Une attention particulière est accordée au traitement de différentes questions, telles que l'automatisation, les informations manquantes (déterminées par des passages satellitaires, des effets de réflexion des nuages, des ombres ou encore la présence de neige), l'apprentissage et les données de validation limitées, les échantillonnages temporels irréguliers (différent nombre d'images disponible pour chaque période et région, données inégalement réparties dans le temps), la variabilité des données, et enfin la possibilité de travailler avec différents ensembles de données et nomenclatures. / The new generation Earth observation missions such as Sentinel-2 (a twin-satellite initiative prepared by the European Space Agency, ESA, in the frame of the Copernicus programme, previously known as Global Monitoring for Environment and Security or GMES) and Venµs, jointly developed by the French Space Agency (Centre National d'Études Spatiales, CNES) and the Israeli Space Agency (ISA), will revolutionize present-day environmental monitoring with the yielding of unseen volumes of data in terms of spectral richness, temporal revisit and spatial resolution. Venµs will deliver images in 12 spectral bands from 412 to 910 nm, a repetitivity of 2 days, and a spatial resolution of 10 m; the twin Sentinel-2 satellites will provide coverage in 13 spectral bands from 443 to 2200 nm, with a repetitivity of 5 days, and spatial resolutions of 10 to 60m. The efficient production of land cover maps based on the exploitation of such volumes of information for large areas is challenging both in terms of processing costs and data variability. In general, conventional methods either make use of supervised approaches (too costly in terms of manual work for large areas), target specialised local models for precise problem areas (not applicable to other terrains or applications), or include complex physical models with inhibitory processing costs. These existent present-day approaches are thus inefficient for the exploitation of the new type of data that the new missions will provide, and a need arises for the implementation of accurate, fast and minimally supervised methods that allow for generalisation to large scale areas with high resolutions. In order to allow for the exploitation of the previously described volumes of data, the objective of this thesis is the conception, design, and validation of a fully automatic approach that allows the estimation of large-area land cover with high spatial, spectral and temporal resolution Earth observation imagery, being generalisable to different landscapes, and offering operational computation times with simulated satellite data sets, in preparation of the coming missions.

<div> <div> <div> <p>An RGBZ synthetic dataset consisting of five object classes in a variety of virtual environments and orientations was combined with a small sample of real-world image data and used to train the Mask R-CNN (MR-CNN) architecture in a variety of configurations. When the MR-CNN architecture was initialized with MS COCO weights and the heads were trained with a mix of synthetic data and real world data, F1 scores improved in four of the five classes: The average maximum F1-score of all classes and all epochs for the networks trained with synthetic data is F1∗ = 0.91, compared to F1 = 0.89 for the networks trained exclusively with real data, and the standard deviation of the maximum mean F1-score for synthetically trained networks is σ∗ _F1= 0.015, compared to σF 1 = 0.020 for the networks trained exclusively with real data. Various backgrounds in synthetic data were shown to have negligible impact on F1 scores, opening the door to abstract backgrounds and minimizing the need for intensive synthetic data fabrication. When the MR-CNN architecture was initialized with MS COCO weights and depth data was included in the training data, the net- work was shown to rely heavily on the initial convolutional input to feed features into the network, the image depth channel was shown to influence mask generation, and the image color channels were shown to influence object classification. A set of latent variables for a subset of the synthetic datatset was generated with a Variational Autoencoder then analyzed using Principle Component Analysis and Uniform Manifold Projection and Approximation (UMAP). The UMAP analysis showed no meaningful distinction between real-world and synthetic data, and a small bias towards clustering based on image background. </p></div></div></div>