Дисертації з теми "Vegetation support"

Research in Virtual Heritage has gained popularity in recent years. Efforts by the community of Virtual Heritage researchers to reconstruct sites considered worthy of preservation span from the historical “built environment”, including the Pyramids at Ghiza and Virtual Reality Notre Dame, to natural heritage sites such as Australia’s Great Barrier Reef and the Virtual Everglades at Florida. Other important efforts to conserve artefacts and educate visitors include Virtual Stonehenge, Pompeii and the Caves of Lascaux. Entire villages, cities and even caves have been constructed as part of virtual conservation efforts. These digital reconstructions have, to date, contributed significant awareness and interest among the general public, providing educational benefits to schoolchildren and new research opportunities to archaeologists and conservationists, to mention but two groups of beneficiaries. Today, to paraphrase the work of Professor Robert J. Stone, Virtual Heritage strives to deliver to a global audience, computer-based reconstructions of artefacts, sites and actors of historic, artistic, religious and cultural heritage in such a way as to provide formative educational experience through the manipulations of time and space. It is realised that the user experience and educational value of a Virtual Heritage site is crucial – the process of virtual reconstruction is as important as its outcome. The total experience therefore, hinges on the modelling accuracy, scientific credibility, and the interactive visualisation capability of a virtual site. However, many interactive media implementations in Virtual Heritage in the recent past have failed to make full use of the advanced interactive visualisation techniques available to researchers. In particular, an element that many end users might consider essential, namely the inclusion of “living” and responsive virtual agents are noticeably lacking in most all Virtual Heritage examples. The addition of these ‘living’ entities and environments could give Virtual Heritage applications a richer, more evolvable content, and a higher level of interactivity. Artificial Life (alife), an emerging research area dealing with the study of synthetic systems that exhibit behaviours characteristic of natural living systems, offers great potential in overcoming this missing element in current Virtual Heritage applications. The present research investigates the feasibility of constructing models of vegetation, exploiting new developments in Artificial Life implemented within a controlled Virtual Environment for application in the field of Archaeology. The specific area of study is the recently discovered and recently named Shotton river valley off the eastern coast of the United Kingdom – a region that once flourished during the Mesolithic Era prior to the post-glacial flooding of the North Sea.

Doctor of Philosophy
Department of Horticulture, Forestry, and Recreation Resources
Charles J. Barden
Recurrent, landscape-level fires played an integral part in the development and persistence of eastern oak (Quercus spp.) forests of the United States. These periodic surface fires helped secure a competitive position for oaks in the regeneration pool by maintaining a desirable species composition and forest structure. This historical fire regime was altered with the European settlement of North America, and fire suppression within forestlands became a standard practice since 1930s. With decades of fire suppression, mature oak-dominated woodlands have widely converted to shade-tolerant tree species. Prescribed fire has successfully been used to enhance oak regeneration in eastern forests. However, oak woodland restoration within the forest-prairie ecotone of the Central plains has not been systematically studied. Fuel beds under shade-tolerant species are often less conducive to fire. Therefore, monitoring fuel loading (FL) and its changes are essential to inform management decisions in an oak regeneration project. Rapid expansion of eastern redcedar (Juniperus virginiana/ERC) is another ecological issue faced by land managers throughout North America’s midcontinent forest-prairie ecotone. Hence, it is worthy to monitor ERC expansion and effects on deciduous forests, to inform oak ecosystem restoration interventions within this region. Therefore, the main objectives of this dissertation were three-fold: (1) understand the effects of prescribed burning and mechanical thinning to encourage oak regeneration; (2) investigate the initial effects of an oak regeneration effort with prescribed fire and mechanical thinning on FL; and (3) monitor the spatio-temporal dynamics of ERC expansion in the forest-prairie ecotone of Kansas, and understand its effects on deciduous forests. The first two studies were conducted on a 90-acre oak dominated woodland, north of Manhattan, Kansas. The experimental design was a 2 (burn) x 2 (thin) factorial in a repeated measures design. The design structure allowed four treatment combinations: burn only (B), thin only (T), burn and thin combined (BT), and a control (C). Burning and thinning treatments were administered in spring 2015. Changes in the FL estimates after the burn treatment revealed that the BT treatment combination consumed more fuel and burned more intensely compared to the B treatment. This observation was reflected in vegetation responses. The thinning reduced the canopy cover significantly, but under enhanced light environments, both oaks and competitive species thrived when no burn was incorporated. In contrast, burn treatments controlled the competitive vegetation. Hence, the most promising results were obtained when both fire and thinning were utilized. The remote sensing study documented the expansion of ERC in three areas of eastern Kansas over 30 years. The use of multi-seasonal layer-stacks with a Support Vector Machines (SVM) supervised classification was found to be the most effective approach to map ERC distribution. Total ERC cover increased by more than 6000 acres in all three study areas investigated in this study between 1986 and 2017. Much of the ERC expansion was into deciduous woodlands. Therefore, ERC control measures should be incorporated into oak woodland restoration efforts within the forest-prairie ecotone of Kansas.

Synoptic global remote sensing provides a multitude of land surface state variables. The continuous collection, for more than 30 years, of global observations has contributed to the creation of a unique and long term satellite imagery archive from different sensors. These records have become an invaluable source of data for many environmental and global change related studies. The problem, however, is that they are not readily available for use in research and application environment and require multiple preprocessing. Here, we looked at the daily global data records from the Advanced Very High Resolution Radiometer (AVHRR) and the Moderate Resolution Imaging Spectroradiometer (MODIS), two of the most widely available and used datasets, with the objective of assessing their quality and suitability to support studies dealing with global trends and changes at the land surface. Findings show that clouds are the major data quality inhibitors, and that the MODIS cloud masking algorithm performs better than the AVHRR. Results show that areas of high ecological importance, like the Amazon, are most prone to lack of data due to cloud cover and aerosols leading to extended periods of time with no useful data, sometimes months. While the standard approach to these challenges has been compositing of daily images to generate a representative map over a preset time periods, our results indicate that preset compositing is not the optimal solution and a hybrid location dependent method that preserves the high frequency of these observations over the areas where clouds are not as prevalent works better. Using this data quality information the Vegetation Index and Phenology (VIP) Laboratory at The University of Arizona produced over 30 years of seamless sensor independent record of vegetation indices and land surface phenology metrics. These data records consist of 0.05-degree resolution global images for daily, 7-days, 15-days and monthly temporal frequency. These sort of remote sensing based products are normally made available through the internet by large data centers, like the Land Processes Distributed Active Archive Center (LP DAAC), however, in this project an online tool, the VIP Data Explorer, was developed to support the visualization, exploration, and distribution of these Earth Science Data Records (ESDRs) keeping it closer to the data generation center which provides a more active data support and distribution model. This web application has made it possible for users to explore and evaluate the products suite before download and use.

Les grandes villes, en periode estivale, developpent de plus en plus frequemment certains problemes lies au phenomenes d'ilot de chaleur urbain, comme les pics de pollution et la surconsommation energetique due a la climatisation. La mineralisation des villes, en remplacant la vegetation et les zones humides par du beton et de l'asphalte contribue a ces nuisances. Notre objectif est alors de determiner l'impact de la vegetation et des bassins ou jets d'eau sur le microclimat urbain et sur les situations de confort thermique ressenties par un individu. Cette etude fait appel aux techniques de modelisation numerique. Dans la premiere partie, une importante synthese bibliographique permet de faire le point sur des domaines aussi varies que la microclimatologie urbaine, la simulation, l'urbanisme, l'arboriculture et le confort thermique en espace exterieur. Ces informations sont utiles lors de la mise en uvre et de l'interpretation des simulations envisagees. Dans la deuxieme partie, nous detaillons la realisation d'un couplage thermo-aeraulique, base sur deux outils numeriques, solene (logiciel d'ensoleillement et de thermique developpe par le cerma) et n3s (code de mecanique des fluides, developpe par edf). Ce couplage necessite le developpement de programmes specifiques et de procedures d'interfacage. Des elements de validation sur des etudes de cas referencees, ainsi qu'une comparaison avec des mesures in-situ sont presentes. L'application de cette demarche a une etude de cas urbaine, la place du millenaire du quartier antigone a montpellier, permet d'analyser l'impact de la vegetation sur le microclimat urbain et les conditions de confort en espace exterieur. La comparaison de trois situations, l'une sans vegetation, l'autre avec la vegetation actuelle et la derniere avec une vegetation a taille adulte, demontre une evolution notable au cours de annees des conditions microclimatiques et une amelioration sensible des situations de confort.

Les approches de requalification de sites dégradés répondent de plus en plus à un double enjeu de dépollution et de valorisation. Les sols de ces sites, dits « sols très anthropisés », sont très souvent caractérisés par un fonctionnement dégradé lié aux processus anthropo-génétiques à l'œuvre dans ces contextes. Ces processus sont à l'origine de la présence dans ces sols de matériaux exogènes technogéniques ou naturels en quantités et natures variées. Dans ce contexte, les études et réflexions sur leur gestion s'effectuent classiquement sur la fraction fine, obtenue après tamisage à 2 mm, et laisse de côté la fraction dite « grossière », souvent surreprésentée (jusqu'à 50-60 % en masse). Or cette fraction, souvent considérée comme inerte, ne l'est pas nécessairement et sa présence peut éventuellement orienter le fonctionnement des écosystèmes et leur gestion future. Une approche méthodologique a alors été développée grâce à laquelle les dimensions et la nature des objets à étudier ont été précisément définis à travers l'utilisation d'une échelle granulométrique adaptée et la construction d'une typologie. Le couplage de cette approche à des caractérisations physico-chimiques a démontré la capacité de discrimination de la stratégie de tri mise en œuvre. La typologie a ainsi pu être amendées avec les informations liées aux composés caractéristiques des catégories de matériaux identifiées. Sur cette base, cinq matériaux modèles (anthracite, brique, calcaire, coke de pétrole et laitier de haut fourneau) ont été sélectionnés afin d'aider à la généralisation des résultats. Un processus d'adaptation des méthodes standardisées d'analyses de sol a alors été développé afin d'évaluer la réactivité des fractions granulométriques grossières. Ces dernières ne sont pas inertes et leur réactivité, bien qu'elle soit réduite par rapport à la fraction fine d'un point de vue massique, est uniquement dépendante du type de matériau et de la surface des particules. Pris seuls, les matériaux modèles sélectionnés présentent un niveau de fertilité insuffisant pour le développement du végétal et ne sont que marginalement à l'origine d'une toxicité. Ces résultats indiquent, qu'au sein du sol, les fractions grossières constituent un stock pérenne d'éléments mobilisables dans le temps à travers le processus d'altération mis en jeu par la pédogenèse. Ceci a été confirmé avec des essais de culture de végétaux sur des substrats à la formulation originale associant fractions grossières, fractions fines et matériaux inertes en conditions contrôlées en laboratoire et en lysimètres à l'échelle pilote de terrain. La contribution directe des fractions grossières à l'alimentation de la solution du sol en éléments puis à la nutrition minérale des végétaux est mise en évidence à court et à moyen termes. Ces fractions jouent également un rôle majeur dans le fonctionnement hydrique du sol en contribuant significativement à la rétention, au stockage de l'eau et à la libération d'éléments. Le type de matériau apparaît être le facteur principal gouvernant l'intensité de ces effets. Ces travaux de thèse ont ainsi permis de replacer les fractions granulométriques grossières dans un rôle actif dans le fonctionnement et l'évolution des sols très anthropisés. Les résultats obtenus apportent des connaissances nouvelles pour le génie pédologique et les acteurs de la gestion des sols très anthropisés. A cet effet, une procédure systématique de diagnostic des fractions granulométriques grossières a été proposée dont les effets doivent servir deux objectifs, i) enrichir les connaissances liées aux constituants des fractions grossières et ii) permettre une prise de décision éclairée, basée sur une connaissance plus complète du contenu d'un sol très anthropisé, autorisant une meilleure gestion en fonction du type de renaturation visée
The approaches of requalification of degraded sites increasingly answer the double challenge of depollution and valorization. The soils of these sites, known as "highly anthropized soils", are often characterized by a disturbed functioning due to anthropogenetic impacts. In particular, anthropic actives are responsible for the presence in these soils of exogenous materials, from both technogenic or natural origins, in varying quantities and natures. In this context, all the diagnoses and considerations regarding their management are carried out on the fine fraction, obtained after sieving at 2 mm, and ignore the so-called "coarse" fraction, which is often overrepresented (up to 50-60% by mass). This fraction, considered as inert, may not necessarily be so and its presence may have important consequences on the functioning of ecosystems and their future management.A dedicated methodological approach was developed. First, the dimensions and nature of the objects to be studied were precisely defined using an appropriate granulometric scale and the construction of a typology. This approach coupled to physico-chemical characterizations confirmed the efficiency of the tested screening strategy and fully enabled a precise and meaningful discrimination among coarse materials. The typology was thus implemented with composition characteristics of the material categories. On this basis, five reference materials (anthracite, brick, limestone, petroleum coke and blast furnace slag) were selected in order to allow for the generalization of the results. An adaptation process of standard soil analysis methods was then developed in order to assess the reactivity of coarse granulometric fractions. Results showed that coarse fractions are not inert. Their reactivity, although reduced compared to the fine fraction from a mass point of view, is only dependent on the type of material and the surface of the particles. Taken alone, the selected model materials present an insufficient level of fertility for plant development and are only marginally at the origin of toxicity. In an overall soil system, the coarse fraction constitutes a perennial stock of elements of interest that can be mobilized over time through the weathering processes generated by pedogenesis. The direct contribution of coarse fractions to plant nutrition was confirmed both under controlled conditions in the laboratory and in lysimeters at the pilot field scale. They also significantly contribute to water retention and storage. The type of material is the main factor governing the intensity of these effects.This thesis work has thus positioned the coarse granulometric fractions as being an active part of the functioning of highly anthropized soils. The results obtained provide new knowledge for stakeholders involved in soil engineering and highly anthropized soils management. To this end, a systematic procedure of diagnosis of coarse granulometric fractions has been proposed, the effects of which should serve two purposes, i) to enrich the knowledge related to the constituents of coarse fractions and ii) to allow an insightful decision-making process, based on a complete knowledge of the reality of a highly anthropized soil, authorizing an optimal and sensible management according to the targeted type of renaturation

Adding a surface rock layer (also called rock armor or rock mulch) to constructed slopes improves erosion resistance but has had mixed effects on revegetation. This study investigated the effects of rock layer depth (no rocks, 10-, 15-, and 20-cm rock layers) and rock size (5-20cm diameter rocks) on vegetation cover. Seeding was applied four times in the first 2 years. After 3 years, plots with a rock layer averaged 7% vegetative cover compared to 85% on plots without a rock layer. There was a nonsignificant trend toward less vegetation with a deeper rock layer.

Precision agriculture requires automated systems for weed detection as weeds compete with the crop for water, nutrients, and light. The purpose of this study is to investigate the use of machine learning methods to classify weeds/crops in agriculture. Statistical methods, support vector machines, convolutional neural networks (CNNs) are introduced, investigated and optimized as classifiers to provide high accuracy at high vehicular speed for weed detection. Initially, Support Vector Machine (SVM) algorithms are developed for weed-crop discrimination and their accuracies are compared with a conventional data-aggregation method based on the evaluation of discrete Normalised Difference Vegetation Indices (NDVIs) at two different wavelengths. The results of this work show that the discrimination performance of the Gaussian kernel SVM algorithm, with either raw reflected intensities or NDVI values being used as inputs, provides better discrimination accuracy than the conventional discrete NDVI-based aggregation algorithm. Then, we investigate a fast statistical method for CNN parameter optimization, which can be applied in many CNN applications and provides more explainable results. This study specifically applies Taguchi based experimental designs for network optimization in a basic network, a simplified inception network and a simplified Resnet network, and conducts a comparison analysis to assess their respective performance and then to select the hyper parameters and networks that facilitate faster training and provide better accuracy. Results show that, for all investigated CNN architectures, there is a measurable improvement in accuracy in comparison with un-optimized CNNs, and that the Inception network yields the highest improvement (~ 6%) in accuracy compared to simple CNN (~ 5%) and Resnet CNN counterparts (~ 2%). Aimed at achieving weed-crop classification in real-time at high speeds, while maintaining high accuracy, the algorithms are uploaded on both a small embedded NVIDIA Jetson TX1 board for real-time precision agricultural applications, and a larger high throughput GeForce GTX 1080Ti board for aerial crop analysis applications. Experimental results show that for a simplified CNN algorithm implemented on a Jetson TX1 board, an improvement in detection speed of thirty times (60 km/hr) can be achieved by using spectral reflectance data rather than imaging data. Furthermore, with an Inception algorithm implemented on a GeForce GTX 1080Ti board for aerial weed detection, an improvement in detection speed of 11 times (~2300 km/hr) can be achieved, while maintaining an adequate detection accuracy above 80%. These high speeds are attained by reducing the data size, choosing spectral components with high information contents at lower resolution, pre-processing efficiently, optimizing the deep learning networks through the use of simplified faster networks for feature detection and classification, and optimizing computational power with available power and embedded resources, to identify the best fit hardware platforms.

The thesis deals with a new hotel construction in Brno. The paper aims to elaborate a project documentation for the construction of the building. It is a detached building in a slightly sloping terrain. The building has three floors and one underground floor. On the underground floor there are jointly-shared garages and technical background of the hotel. On the first floor there is a reception, a restaurant, a café and an open-air terrace. The second and the third floor is designed to accommodate guests in double rooms and suites. There is also a manager´s office and a conference room. The bearing construction consists of a ferroconcrete skeleton with a bricked outside envelope of aerated concrete blocks Ytong. The skeletal construction is built on a monolithic foundation footings. The building is covered in a flat single-coat roof which is partly designed as a vegetation roof. The facade of the building is glazed in the northern part of the 1st floor and some parts of the facade are coloured green, the others are of various colours. The building is insulated with a contact insulation system.

Thesis on "Planetarium with coffee" is processed in the form of project documentation for construction. The proposed facility is designed as a two-storeyed without basement with the main ground plan dimensions of 52,8 x 27,47 m and is located near the village Lošov close to Olomouc. The building is operationally divided into a part intended for public entrance and a part for operation and maintenance. Loadbearing structural system is reinforced concrete monolithic skeleton based on monolithic reinforced concrete foundation footings. Ceiling structure consists of reinforced concrete, monolithic, locally supported, cross reinforced floor slab. Vertical external walls are made of ceramic blocks with contact thermal insulation system. Part of the front rounded facade is designed as glass fronted. The building is covered by a warm flat roof. Roof cladding consists of vegetation layers, walkable terrace or non-walkable structure of roof cladding. The planetarium itself is covered by a reinforced concrete dome with a thermal insulation layer of plates of foamed glass. Fillings of doors and windows are made of aluminum profiles with isolating triple glazing. The building meets the requirements for barrier-free use of buildings.

This diploma thesis deals with the design and elaboration process of a project documentation of medical center in city Hlinsko. New building is situated in the northern part of the city in the area determined for constructions of public infrastructure. It´s a four-floor object with partial basement and slant mansard and flat vegetative roof.Object is based on foundation strips and footings made from reinforced concrete.The bearing and internal walls are designed from KALKSANDSTEIN lime sand blocks.The beams and collums are used in the placed designed with open disposition.The ceiling in 1.PP is designed as reinforced concrete for securing fire safety.The rest od ceiling constructions is designed from prestressed concrete panels SPIROLL,with exception in 4.NP,where the ceiling constructions are made from solid roof system YTONG KOMFORT,which is layed on frame system made of steel and reinforced concrete.The peripheral walls of 1.PP are made from lost formwork, which is insulated with extruded polystyrene.The walls in floors above ground are insulated with combination of external thermal insulation system (ETICS) and ventilated facade with CEMBRIT SOLID and PATINA facade cladding panels. The building is functionally divided into several part.Cafe, pharmacy, ambulance and medical center, each with it´s own entrance.Pharmacy is equipped with shop, medicine mixer and storage.Ambulance is equipped with garage slots, background for employees, room of crisis preparedness and operating center.Medical centrum is equipped with emergency, RTG, operating theater, rehabilitations and other types of examination rooms.Each examination room is equipped with waiting room.For more comfort the center has a background for employees.The whole object is designed with forced air exchange. Designed object is barrier-free.In front of the building is situated parking slots for cars, motorbikes and ambulances.Total number of slots is 74 places, which 5 is designed for disabled people.

A comercialização de produtos florestais não madeireiros (PFNMs) popularizou-se como atividade promotora do desenvolvimento socioeconômico de comunidades florestais com baixo impacto ambiental. Há evidências, contudo, de que a exploração possa produzir efeitos ecológicos e socioeconômicos negativos, sugerindo que é necessário monitorar tais iniciativas. A comercialização frequentemente ocorre em áreas remotas e em contextos de pobreza, como é o caso de muitas daquelas da Amazônia brasileira. Desta forma, é necessário que o monitoramento restrinja-se a avaliar os efeitos evidenciados como mais comuns em estudos anteriores. Este estudo revisa e sintetiza as evidências científicas dos efeitos da exploração de PFNMs sobre parâmetros ecológicos e socioeconômicos e, a partir daí, indica aqueles mais relevantes ao monitoramento. O estudo também levanta até que ponto o monitoramento é implementado no contexto da Amazônia brasileira e avalia quais os parâmetros importantes e viáveis de monitoramento neste caso específico. Para isso, foram realizadas revisões sistemáticas da literatura e a consulta a profissionais da área através do método Delphi. Os resultados indicam que efeitos ecológicos negativos são frequentes, principalmente quando se coletam folhas ou cascas. Alterações em órgãos ou processos fisiológicos e a taxa de sobrevivência dos espécimes explorados são parâmetros que devem ser monitorados, em particular quando se coletam frutos e partes vegetativas. Para todos os tipos de PFNMs, o tamanho e a estrutura populacional são parâmetros prioritários ao monitoramento. A riqueza de espécies da comunidade explorada merece atenção, principalmente quando se coletam frutos. Para os aspectos socioeconômicos, efeitos positivos foram mais frequentes que negativos. A contribuição da renda monetária obtida com o comércio na renda total, a regularidade de ingresso desta renda e o papel dos PFNMs como recursos de salvaguarda são parâmetros do capital financeiro prioritários ao monitoramento. Para o capital social, o empoderamento feminino, a coesão de grupo e o acesso aos benefícios gerados pela comercialização devem ser monitorados. Na Amazônia brasileira são poucas as iniciativas de implementação do monitoramento da comercialização de PFNMs. Embora este seja considerado importante, existem dificuldades que derivam principalmente da falta de apoio institucional, políticas de incentivo e de recursos financeiros. Os profissionais participantes do Delphi consideram que os parâmetros ecológicos mais importantes a monitorar neste contexto são o tamanho e a estrutura populacional do recurso explorado, o aumento da taxa de mortalidade, a quantidade total de recurso extraída e a técnica de coleta utilizada. Para os parâmetros econômicos, aspectos do mercado, como o preço pago ao coletor, a demanda e a qualidade do produto, bem como a renda monetária obtida pelos indivíduos são os parâmetros considerados mais importantes. Por fim, para os aspectos sociais, os efeitos na cultura, na qualidade de vida e na organização interna da comunidade foram priorizados. Os profissionais indicam que é viável estabelecer o monitoramento dos parâmetros levantados.
Amazon, non-timber forest products, ecological effects, socioeconomic effects, monitoring.

Dissertação de Mestrado Integrado em Engenharia Electrotécnica e de Computadores apresentada à Faculdade de Ciências e Tecnologia
Nos últimos anos, os Veículos Aéreos Não Tripulados (VANT) têm sido cada vez mais utilizados quer para fins recreativos, quer para fins académicos. Até agora, imagens aéreas eram apenas recolhidas por satélite ou através de voos de elevada altitude, o que tornava o processo economicamente inviável.Os VANT, aliados com vários sensores, constituem uma excelente via para realizar monitorização à distância. O principal objetivo deste trabalho é classificar vários tipos de árvores em imagens aéreas recolhidas por um VANT. Para tal, foi usada uma plataforma aérea quadrotor com uma câmara de ação convencional para recolher imagens de várias espécies de árvores. A deteção e classificação das árvores nas imagens foi obtida através de um detetor de objetos baseado em descritores de canais agregados e em Máquinas de Vetores de Suporte (MVS). O detetor é aplicado a cada uma das imagens, identificando cada uma das regiões de interesse. Para cada uma dessas deteções, 4096 descritores são extraídos com um Rede Neuronal Convolucional (RNC), que são posteriormente usados como entrada para a MSV.Foram criados dois conjuntos de imagens e foram realizados testes a cada uma das componentes do sistema, individualmente, bem como à conjugação de todos os métodos.O primeiro conjunto de dados é usado para testar o detetor e o classificador separadamente e, por fim, os dois conjuntos são usados para testar a totalidade do sistema. O detetor gerou valores de exatidão na ordem dos 60% enquanto que o classificador gerou valores na ordem dos 77%. Por fim, o sistema resultante da conjugação do detetor e do classificador originou valores na ordem dos 65% usando o primeiro conjunto e 59% usando o segundo conjunto de imagens.
In the last years, Unmanned Aerial Vehicles (UAVs) have been used more and more with recreational purposes as well as academic ones. So far, aerial imagery was only obtained by satellite or by high altitude flights which are expensive to obtain, making the process, sometimes, economically unfeasible.The UAVs, associated with various types of sensors, constitute an excellent way to achieve advanced monitoring at a distance. The main goal of this dissertation is to classify various types of trees on aerial images. To achieve that purpose, a quadrotor equipped with a conventional action camera was used to retrieve images containing various tree species. The tree detection and classification was achieved by an Aggregated Channel Features (ACFs) Object Detector and Support Vector Machine (SVM). The detector is applied to each image, identifying every Region of Interest (ROI). For each one ofthose ROIs, 4096 features are extracted with a pre-trained Convolutional Neural Network (CNN), which are posteriorly used as input of the SVM.Two diferent datasets were created and several tests were made to each system componente individually as well as the totality of the system. First, dataset #1 is used to test both detector and classifier separately and, then, the overall system is tested using the two datasets. The detector revealed 60% of accuracy and the classifier originated values of accuracy around 77%. Finally, the resultant system has an overall accuracy of 65% using dataset #1 and 59% using dataset #2. .

Dissertação de Mestrado Integrado em Engenharia Electrotécnica e de Computadores apresentada à Faculdade de Ciências e Tecnologia
Na Agricultura de Precisão (AP), a detecção e classificação de vegetação em culturas herbáceas (e.g., vinhas e árvores) é um passo crucial para a definição de objectivos consequentes, tais como a utilização de diferentes fertilizantes ou distintos níveis de hidratação.Um Veículo Aéreo Não Tripulado (VANT) foi montado e testado com o objetivo de desenvolver uma ferramenta de classificação (câmara ``multiespectral'') para ser usada em diferentes culturas. O sistema compreende uma plataforma aérea ``quad-rotor'', capaz de voar até alturas de 100 m acima do nível do solo. O presente trabalho desenvolve diferentes abordagens de segmentação que resultam na identificação da Região de Interesse (RDI) e, posteriormente, na sua classificação de acordo com a previsão fornecida por dois classificadores: Máquinas de Vectores de Suporte (MVS) e Árvores de Decisão (AD). As imagens são capturadas por um sistema composto por uma câmara (NoIR Raspberry Pi (RPi) - com um conjunto de cinco filtros ópticos acoplados à sua lente) conectada a um RPi, colocado no VANT para detecção de vegetação. As imagens foram adquiridas através de voos sobre os seguintes campos agrícolas: três vinhas com diferentes variedades de casta e florestas de cinco espécies de árvores (eucaliptos, pinheiros, oliveiras, laranjeiras e magnólias). Os testes mostram o desempenho dos dois classificadores, de acordo com a RDI identificada previamente por um algoritmo baseado em ``thresholding'' através de valores fornecidos pelo índice de vegetação NDVI. Os dois classificadores recebem dados de entrada fornecidos pelo cálculo de oito índices de vegetação. O algoritmo gerou valores de exactidão de 72% e 73% (para o Sistema de Reconhecimento de Padrões (SRP) associado ao conjuntos de dados #1 e #2, respectivamente, relacionados com as MVS) e 74% e 79% (para o SRP de acordo com os conjuntos de dados #1 e #2, respectivamente, relacionadas com os AD). O algoritmo é totalmente automatizado e a classificação é fornecida a partir do RPi para uma base de controlo, em tempo real, através duma conexão por Wi-Fi.
In Precision Agriculture (PA), detecting and classifying the vegetation in herbaceous crops (e.g., vineyards and trees) is a crucial step prior to address further objectives, such as specifying either different fertilizers or distinguished hydration levels. An Unmanned Aerial Vehicle (UAV) was assembled and tested with the aim of developing a classification tool (``multispectral'' camera) for both types of vineyard and tree species. The system comprises a quad-rotor aerial platform capable of flying up to heights of 100 m above the ground level. The present research work develops different segmentation approaches which result in the identification of the Region of Interest (ROI) and afterwards their classification according to the prediction provided by two different machine learning classifiers: Support Vector Machines (SVM)s and Decision Trees (DT)s. Along with the general description of the procedure, remotely-sensed images captured with a sensor (a NoIR Raspberry Pi (RPi) camera with a five optical filters wheel attached) connected to an RPi and mounted on the UAV, were applied for vegetation detection. Images were acquired while hovering above both fields of three vineyard species (with different grape varieties) and forests of five tree species (eucalyptus, pine trees, olive trees, orange trees and magnolias). The tests show the performance of both classifiers, according to the ROI identified by a thresholding algorithm based on Normalized Difference Vegetation Index (NDVI) measurements. The two classifiers receive input data provided by the computation of eight Vegetation Indices (VI)s.The algorithm has led to accuracy values of 72\% and 73\% (for the Pattern Recognition System (PRS) according to datasets #1 and #2, respectively, related to the SVMs and 74% and 79% (for the PRS according to datasets #1 and #2, respectively, related to the DTs). The entire algorithm is totally automated and the classification output is provided from the RPi to a ground station in real-time, by a Wi-Fi socket connection.

Meadow vegetation in the Krkonoše Mountains National Park is classified in this master thesis using aerial hyperspectral data from sensor AISA and Support Vector Machines (SVM) and Neural Networks (NN) classification algorithms. The main goals of the master thesis are to determine the best settings of SVM parameters and to propose an ideal design for a training dataset for this classification algorithm and mapping of the meadows in the Krkonoše mountains. The criterion of the tests will be the result of classification accuracy (confusion matrices and kappa coefficient). The additional goal of the master thesis is to compare performances of both utilized classifiers, especially regarding the amount of training pixels necessary for successful classification of the mountainous meadow vegetation. Classification maps of the area of interest and Python scripts are the main outputs of the master thesis. These outputs will be handed over to the Administration of the Krkonoše Mountains National Park for further utilization in the monitoring and protecting these valuable meadow vegetation communities. Key words: hyperspectral data, AISA, Support Vector Machines, Neural Networks, training dataset, mountainous meadow vegetation

Promieniowanie elektromagnetyczne interferuje w sposób selektywny z molekułami materii, tworząc charakterystyki spektralne. Są one unikatowe dla poszczególnych gatunków czy zbiorowisk roślinnych, odzwierciedlając ich stan fizjologiczny, budowę anatomiczną i morfologiczną. Są one analizowane przy użyciu teledetekcji hiperspektralnej, która dzięki wysokiej rozdzielczości spektralnej i radiometrycznej, a w przypadku obrazów lotniczych także i przestrzennej pozwalają na identyfikację zbiorowisk roślinnych. Celem niniejszej pracy była klasyfikacja roślinności subalpejskiej i alpejskiej Karkonoszy na podstawie lotniczych obrazów hiperspektralnych APEX. Do tego celu posłużono się jedną z najbardziej obiecujących metod, jaką są maszyny wektorów nośnych (Support Vector Machine, SVM). Obszar badań obejmował najwyższe partie polskich oraz czeskich Karkonoszy w granicach parków narodowych, na których występowały zbiorowiska roślinności subalpejskiej i alpejskiej. Obrazy APEX składające się z 288 kanałów spektralnych w zakresie 380,5-2500 nm o rozdzielczości przestrzennej 3,12 m pozyskane zostały 10 września 2012 roku przez Niemiecką Agencję Kosmiczną (DLR) w ramach projektu European Facility for Airborne Research (EUFAR) HyMountEcos (Hyperspectral Remote Sensing for Mountain Ecosystem). Obrazy APEX zostały skalibrowane radiometrycznie oraz skorygowane geometrycznie i atmosferycznie w belgijskim instytucie VITO (operator skanera APEX). Następnie obrazy poddano redukcji przestrzeni spektralnej przy użyciu trzech metod: Principal Component Analysis, Minimum Noise Fraction oraz wyborze najbardziej informacyjnych kanałów spektralnych. Jako materiał referencyjny do klasyfikacji zbiorowisk roślinnych wykorzystana została mapa roślinności nieleśnej w skali 1:10 000 opracowana przez B. Wojtunia i L. Żołnierza (2002), obejmująca 48 zbiorowisk roślinnych należących do 13 typów roślinności. Na podstawie mapy referencyjnej z roku 2002, ortofotomapy o rozdzielczości przestrzennej 12 cm, a także informacji pozyskanych w terenie w sierpniu 2013 oraz 2014 roku z wykorzystaniem odbiornika GPS Trimble GeoXT, opracowane zostały wzorce o klasyfikacji i weryfikacji uzyskanych wyników. Klasyfikacje metodą maszyn wektorów nośnych przeprowadzano na obszarach testowych okolic Szrenicy oraz Małego Stawu, przy użyciu różnych funkcji jądra SVM (liniowej, radialnej, wielomianowej i sigmoidalnej). Modyfikowano także parametry obliczeniowe SVM, a także testowano różne zestawy pikseli treningowych, weryfikacyjnych oraz liczby klas. Finalnie, klasyfikacja przeprowadzona została na całym obszarze Karkonoszy znajdującym się powyżej 1200 m n.p.m., pokrywającym się z linią zobrazowania APEX. W wyniku otrzymano obrazy poklasyfikacyjne w postaci map prezentujących rozkład przestrzenny 24 klas roślinności subalpejskiej i alpejskiej Karkonoszy. Zestawy pikseli weryfikacyjnych pozwoliły na utworzenie macierzy błędów, a także obliczenie dokładności całkowitych, współczynników kappa oraz dokładności charakteryzujących poszczególne klasy: producenta i użytkownika. Dokładność całkowita klasyfikacji całego obszaru Karkonoszy wyniosła 85,5%, kappa 0,84, również dokładności uzyskane dla poszczególnych klas zbiorowisk roślinnych były wysokie – 17 z 24 klas zaklasyfikowało się na poziomie powyżej 80% dokładności producenta. Najwyższe dokładności osiągnięto dla rozległych, homogenicznych klas (murawy halne, murawy bliźniczkowe, traworośla wysokogórskie, zbiorowiska trzęślicy modrej), najniższe dla klas reprezentowanych przez mniejsze zestawy treningowe, mniej jednorodne i trudne do identyfikacji ze względu na niewielkie obszary, najczęściej zacienione (ziołorośle miłosny górskiej, zarośla wierzby lapońskiej). Przedstawione wyniki potwierdzają, że dane hiperspektralne w połączeniu z metodą maszyn wektorów nośnych pozwalają klasyfikować górskie zbiorowiska roślinne.
Plant species and communities have specific properties, which can be analysed using hyperspectral remote sensing. Large number of narrow, contiguous, spectral bands at high spatial and spectral resolution allows to identify and analyse vegetation communities. The aim of the study was to classify subalpine and alpine vegetation communities based on APEX hyperspectral images using Support Vector Machines (SVM). The study area covers the subalpine and alpine vegetation zone of the Giant Mountains in Poland and Czechia. The APEX data characterized by 288 bands in the wavelength range 380.5-2500 nm and the spatial resoultion of 3.12 m were acquired on 10th September 2012 by the DLR Braunschweig (Germany) in the framework of the HyMountEcos (EUFAR) project. APEX data were calibrated radiometrically and corrected geometrically and atmospherically at VITO (Belgium) facilities. After that the dimensionality reduction using three different methods, involving Principal Component Analysis and Minimum Noise Fraction methods and spectral band selection approach, was performed. As reference the vector map of non-forest vegetation distribution at the 1:10 000 scale elaborated by Wojtuń and Żołnierz (2002) and field mapping were used. Based on reference vegetation map, 12-cm resolution orthophotomap and terrain validation performed in 2013 and 2014 using the GPS Trimble GeoXT receiver training and validation patterns were collected. Support Vector Machines were optimised on test areas near Szrenica and Mały Staw using different kernel functions, parameters, training/validation sample sets and different number of classes. Finally, classification was performed on whole area of Giant Mountains above 1200 m a.s.l., covering APEX flightline. As a result, post classification maps of 24 classes of subalpine and alpine vegetation were obtained. Validation samples allowed to assess the total, user, producer accuracies, kappa coefficient and error matrix. Overall accuracy for whole Giant Mountain area was 85.5%, kappa was 0.84, also accuracies for each class of vegetation communities were high – 17/24 classes were classified of more than 80% of producer accuracy. The highest accuracies achieved large, homogeneous classes (Carici (rigidae)-Festucetum airoidis, Carici (rigidae)-Nardetum, Calamagrostion, Molinia caerulea), the lowest – classes represented by smaller training sample sets, less homogeneous and difficult to identify because of the small areas, usually shaded (Adenostyletum alliariae, Salicetum lapponum). The results show that Imaging Spectroscopy and used methods are useful for the classification of the dominant plant communities of mountain areas.

Tese de Doutoramento em Ciências Ambientais.
O principal objectivo desta Tese de Doutoramento foi avaliar a importância e demonstrar o acréscimo de precisão e eficiência associado ao uso de Tecnologias de Informação Geográfica (nomeadamente Detecção Remota e Sistemas de Informação Geográfica) no âmbito de instrumentos de planeamento e gestão dos recursos naturais em “Pequenas Ilhas Oceânicas”. Três diferentes escalas geográficas foram usadas para enquadrar e desenvolver os cinco casos de estudo que constituem os capítulos (e artigos científicos) desta Tese: o Arquipélago dos Acores (escala regional), a ilha de São Miguel (escala de ilha) e a Zona de Proteção Especial Pico da Vara/Ribeira do Guilherme (escala local). Os cinco estudos desenvolvidos foram os seguintes: (1) Proposta metodológica para a definição da Estrutura Ecológica em pequenas ilhas oceânicas: o caso de estudo da ilha de São Miguel (Arquipélago dos Açores, Portugal); (2) Proposta metodológica para a elaboração de Planos de Gestão de Sítios da Rede Natura 2000; (3) Avaliação da adequabilidade de imagens de satélite de alta resolução para cartografia da vegetação em Áreas Protegidas localizadas em pequenas ilhas oceânicas; (4) Detecção e cartografia de espécies exóticas invasoras lenhosas nas Áreas Protegidas dos Acores usando imagens de satélite de alta resolução; (5) Correspondência entre a componente espacial do Programa GMES e as políticas ambientais das Regiões Ultraperiféricas da Europa – o caso de estudo da Região Autónoma dos Acores. Todos os resultados destes estudos reforçaram a pertinência e a importância da produção, integração e processamento de Informação Geográfica de qualidade em todas as fases dos procedimentos de Planeamento e Gestão Territorial, maximizando a sua importante função de suporte operacional a decisão em qualquer escala geográfica (regional, ilha ou local), alicerçada em informação mais precisa e objectiva e com melhor relação custo-benefício, por via do uso adequado das tecnologias de Detecção Remota e Sistemas de Informação Geográfica.
ABSTRACT: The main goal of this Doctoral Thesis was to assess the importance and effectiveness of using Geographic Information and Remote Sensing Technologies within the framework of different Small Islands’ Natural Resources Planning and Management instruments in order to improve their efficiency. Three distinct geographic scales, Azores Archipelago (at regional level), S. Miguel Island (at island level) and Pico da Vara/Ribeira do Guilherme Special Protected Area (at local level) were used as study areas for five different case-studies, individually described and analyzed: (1) Mapping the ecological structure in small islands: the case-study of S. Miguel island (Archipelago of the Azores, Portugal); (2) A Methodological Proposal for the Development of Natura 2000 Sites Management Plans; (3) Assessing the effectiveness of high resolution satellite imagery for vegetation mapping in small islands protected areas; (4) Mapping invasive woody plants in Azores Protected Areas by using high-resolution multispectral imagery; (5) Linking GMES Space Component Sentinel Missions to Outermost Region’s environmental policies development – the Azores (Portugal) case-study. All the case-study’s conclusions have converged and reinforced the importance of using and integrating accurate Geographic Information and Remote Sensing Technologies into all phases of Land Management and Environmental Planning instruments, in order to increase the accuracy of the results and also the cost-effectiveness of the operational schemes.
This Ph.D. Research Project (M3.1.2/F/025/2007) is supported by the Regional Fund for Science and Technology funded by the Azorean Regional Government/Secretary of Science, Technology and Equipment.

碩士
國立政治大學
行政管理碩士學程
97
The 21st century is a time of economic prosperity and technological advancement. However, the world is facing challenges from the problems of the rapidly aging population, increasing longevity, the surging of chronic disease, in addition to epidemics and flues spreading faster then before. Most hospital beds in Taiwan are considered to be “acute hospital beds”, where vegetative patients are considered by the National Health Insurance to be suffering from a chronic disease and not as major illness/injury. There is no extra medical subsidy for these vegetative patients, and only limited payments for home care. Unfortunately, due to the inefficient funding for health insurance, we are concerned that medical resources might become exhausted in the future. Also, a vegetated patient is required to leave the hospital whenever his or her condition becomes stable. They are unable to remain in the hospital for long-term and professional care, which causes the families with vegetated patients to suffer emotionally and economically. Thus, studying ways for vegetated patients to obtain more professional, human, and dignified care is an important issue worthy of research. This study uses an in-depth interview method from quantitative research. 12 families with vegetative patients in Taipei County are the subjects of the interview. In-depth discussions are conducted on the difficulties for families who take care of a vegetative patient, and who have employment in the urban area. The information obtained was categorized and analyzed. 1. Nursing manpower insufficiency Due to the shifting of family structure in the urban area, there are not enough people in the family to take care of the vegetated patient. It is not possible to predict when a medical breakthrough will enable the patient to regain consciousness. Moreover, there is only a slight chance a vegetated patient suffering from head trauma for a long period of time can be revived again. 2. Dependence on foreign nurse aides By hiring foreign nurse aids, the dual-earner families in the Taipei metro area are able to get relief from the burden of caring for a vegetative patient. However, it is not a long-term solution. 3. The need for long-term care. Therefore, it is very important to construct a long-term care system for the vegetated patient to be able to receive quality medical care. Furthermore, Taiwan will benefit from showing respect for life, the importance of social care, and the value of helping each other. 4. Trying new concepts. Perhaps in the foreseeable future, we will accept things such as euthanasia or pre-paid medical contracts due to the changes in education and social attitudes. We might accept that one of our family members might be facing long-term care in the future, so any one of us is able to make arrangements before an unforeseen accident or illness results in becoming a vegetated patient. 5. Co-operation between the government and the people. This is an important issue for the method of constructing a long-lasting system that can combine medicine, labor, social welfare, and benefits for senior citizens and vegetated patients. However, it will take the cooperation of both government and the people to make Taiwan into a peaceful land that will respect their elders and respect life. Some suggestions have been developed as a result of this study: 1. suggestions to families: a balanced diet, exercise regularly, and strengthen the unity between family members. 2. suggestions to the medical team: develop hospital websites, prevent diseases that can cause vegetation, and the social worker consulting team should help families to implement a long-term community care system. 3. Suggestions to schools: add life and death lessons into the school’s curriculum, reinforce the concept of health education and a basic caring method, as well as increase community service, and teach methods for relieving stress. 4. Suggestions to the government: linking information by using a single window to handle all processes, use colloquial language for promoting policies, the policies should first be implemented in off-shore islands before being used in the main island, implement a long-term care system, and provide subsidies to those who hire domestic nurse aids. Keyword： National Health Insurance, Vegetative, Long-Term care, Home care, Euthanasia, Pre-paid medical contract