Dissertations / Theses on the topic 'Aerial photogrammetry – data processing'

This study demonstrates the value of historical aerial photographs as a source for monitoring long-term landslide evolution, which can be unlocked by using appropriate photogrammetric methods. The understanding of landslide mechanisms requires extensive data records; a literature review identified quantitative data on surface movements as a key element for their analysis. It is generally acknowledged that, owing to the flexibility and high degree of automation of modern digital photogrammetric techniques, it is possible to derive detailed quantitative data from aerial photographs. In spite of the relative ease of such techniques, there is only scarce research available on data quality that can be achieved using commonly available material, hence the motivation of this study. In two landslide case-studies (the Mam Tor and East Pentwyn landslides) the different types of products were explored, that can be derived from historical aerial photographs. These products comprised geomorphological maps, automatically derived elevation models (DEMs) and displacement vectors. They proved to be useful and sufficiently accurate for monitoring landslide evolution. Comparison with independent survey data showed good consistency, hence validating the techniques used. A wide range of imagery was used in terms of quality, media and format. Analysis of the combined datasets resulted in improvements to the stochastic model and establishment of a relationship between image ground resolution and data accuracy. Undetected systematic effects provided a limiting constraint to the accuracy of the derived data, but the datasets proved insufficient to quantify each factor individually. An important advancement in digital photogrammetry is image matching, which allows automation of various stages of the working chain. However, it appeared that the radiometric quality of historical images may not always assure good results, both for extracting DEMs and vectors using automatic methods. It can be concluded that the photographic archive can provide invaluable data for landslide studies, when modern photogrammetric techniques are being used. As ever, independent and appropriate checks should always be included in any photogrammetric design.

Unmanned aerial vehicle (UAV) technology was used to determine tidal extent in Kimages Creek, a restored tidal wetland located in Charles City County, Virginia. A Sensefly eBee Real-Time Kinematic UAV equipped with the Sensor Optimized for Drone Applications (SODA) camera (20-megapixel RGB sensor) was flown during a single high and low tide event in Summer 2017. Collectively, over 1,300 images were captured and processed using Pix4D. Horizontal and vertical accuracy of models created using ground control points (GCP) ranged from 0.176 m to 0.363 m. The high tide elevation model was subtracted from the low tide using the ArcMap 10.5.1 raster calculator. The positive difference was displayed to show the portion of high tide that was above the low tide. These results show that UAVs offer numerous spatial and temporal advantages, but further research is needed to determine the best method of GCP placement in areas of similar forest structure.

During the past three years, a research project has been carried out in the Department of Mechanical Engineering at UCT, directed at developing a system to accurately reproduce three-dimensional (3D), sculptured surfaces on a three axis computer numerically controlled (CNC) milling machine. Sculptured surfaces are surfaces that cannot easily be represented mathematically. The project was divided into two parts: the development of an automatic noncontact 3D measuring system, and the development of a milling system capable of machining 30 sculptured surfaces (Back, 1988). The immediate need for such a system exists for the manufacture of medical prostheses. The writer undertook to investigate the measurement system, .with the objective to develop a non-contact measuring system that can be used to 'map' a sculptured surface so that it can be represented by a set of XYZ coordinates in the form required by the milling system developed by Back (1988). This thesis describes the development of a PC-based near-realtime photogrammetry system (PHOENICS) for surf ace capture. The topic is introduced by describing photogrammetric principles as used for non-contact measurements of objects. A number of different algorithms for image target detection, centering and matching is investigated. The approach to image matching adopted was the projection of a regular grid onto the surface with subsequent matching of conjugate grid intersections. A general algorithm which automatically detects crosses on a line and finds their accurate centres was developed. This algorithm was then extended from finding the crosses on a line, to finding all the intersection points of a grid. The algorithms were programmed in TRUE BASIC and specifically adapted for use with PHOENICS as an object point matching tool. The non-contact surface measuring technique which was developed was used in conjunction with the milling system developed by Back (1988) to replicate a test object. This test proved that the combined system is suitable for the manufacture of sculptured surf aces. The accuracy requirements for the manufacture of medical prostheses can be achieved with the combined measuring and milling system. At an object-to-camera distance of 0.5 m, points on a surface can be measured with an accuracy of approximately 0.3 mm at an interval of 5 mm. This corresponds to a relative accuracy of 1:1600. Back (1988) reported an average undercutting error of 0.46 mm for the milling system. This combines to an uncertainty of 0.55 mm. Finally, the limitations of PHOENICS at its prototype stage as a surface measuring tool are discussed, in particular the factors influencing the system's accuracy. PHOENICS is an ongoing project and the thesis is concluded by some recommendations for further research work.

ITC/USA 2011 Conference Proceedings / The Forty-Seventh Annual International Telemetering Conference and Technical Exhibition / October 24-27, 2011 / Bally's Las Vegas, Las Vegas, Nevada
In this paper we investigate the different stages that allow us to create a model that would provide a better understanding of what happens on certain parameters that measure physical quantities related to the behavior of both, burst and reaction, unmanned aircraft as well as unmanned helicopters based on a data transmission to land via radio modem.

Thesis (Ph.D.) - University of Glasgow, 2009.
Ph.D. thesis submitted to the Faculty of Physical Sciences, Department of Geographical and Earth Sciences and the Faculty of Arts, Department of Archaeology, University of Glasgow, 2009. Includes bibliographical references. Print version also available.

Submitted by VANESSA JORDÃO MARCATO FERNANDES null (vanessamarcato@yahoo.com.br) on 2017-01-30T18:15:29Z No. of bitstreams: 1 fernandes_vjm_tese.pdf: 25329126 bytes, checksum: 92c837eb39ae9af74c15ccf278cd2a84 (MD5)
Approved for entry into archive by LUIZA DE MENEZES ROMANETTO (luizamenezes@reitoria.unesp.br) on 2017-02-03T16:39:11Z (GMT) No. of bitstreams: 1 fernandes_vjm_dr_prud.pdf: 25329126 bytes, checksum: 92c837eb39ae9af74c15ccf278cd2a84 (MD5)
Made available in DSpace on 2017-02-03T16:39:11Z (GMT). No. of bitstreams: 1 fernandes_vjm_dr_prud.pdf: 25329126 bytes, checksum: 92c837eb39ae9af74c15ccf278cd2a84 (MD5) Previous issue date: 2016-12-19
Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP)
Esse trabalho propõe o desenvolvimento de um método para a extração automática de contornos de telhados de edifícios com a combinação de dados de Varredura a LASER Aerotransportado (VLA) e dados fotogramétricos e campos aleatórios de Markov (MRF). Inicialmente, um Modelo Digital de Superfície normalizado (MDSn) é gerado através da diferença entre o Modelo Digital de Superfície (MDS) e o Modelo Digital de Terreno (MDT), obtidos a partir da nuvem de pontos LASER. Em seguida, o MDSn é segmentado para a obtenção dos polígonos que representam objetos altos da cena. Esses polígonos são projetados na imagem para restringir o espaço de busca para a segmentação da imagem em regiões. Esse processo possibilita a extração de polígonos na imagem que representem objetos altos. O processo de identificação de contornos de telhados, em meio aos objetos altos detectados na imagem, na etapa anterior, é realizado através da otimização de uma função de energia estabelecida com base em MRF que modela propriedades específicas de contornos de telhados de edifícios. No modelo MRF são utilizados tanto os polígonos extraídos da imagem quanto os extraídos dos dados VLA. A função de energia é otimizada pelo método Algoritmo Genético (AG). O método proposto nesse trabalho foi avaliado com base em dados reais - imagens aéreas de alta resolução e dados VLA. Os resultados obtidos na avaliação experimental mostraram que a metodologia funciona adequadamente na tarefa de extrair os contornos de telhados de edifícios. A função de energia proposta associada ao método de otimização AG diferenciou corretamente os contornos de telhados de edifícios dos demais objetos altos presentes nas cenas. Os contornos de telhados extraídos apresentam boa qualidade, o que é evidenciado por meio dos índices de completeza e correção obtidos pela avaliação numérica. Com base nos índices médios obtidos para cada experimento, têm-se as médias de completeza e correção para os experimentos iguais a 90,96% e 98,99%, respectivamente. Os valores máximos de completeza e correção são de 99,19% e 99,94%, respectivamente, e os valores mínimos de 78,08% e 97,46%, respectivamente. Os menores valores de completeza estão associados às áreas de oclusão por vegetação e presença de sombras.
This paper proposes a method for the automatic extraction of building roof contours through a combination of Airborne Laser Scanner (ALS) and photogrammetric data, and Markov Random Field (MRF). Initially, a normalized digital surface model (nDSM) is generated on the basis of the difference between the digital surface model and the digital terrain model, obtained from the LiDAR point cloud. Then the nDSM is segmented to obtain the polygons representing aboveground objects. These polygons are projected onto image to restrict the search space for image segmentation into regions. This process enables the extraction of polygons in the image representing aboveground objects. Building roof contours are identified from among the aboveground objects in the image by optimizing a Markov-random-field-based energy function that embodies roof contour specific properties. In the MRF model are used both polygons extracted from image and from ALS data. The energy function is optimized by the Genetic Algorithm (GA) method. The method proposed in this work was evaluated based on real data - high-resolution aerial images and ALS data. The results obtained in the experimental evaluation showed that the methodology works adequately in the task of extracting the contours of building roofs. The proposed energy function associated with the GA optimization method correctly differentiated the building roof contours from the other high objects present in the scenes. The extracted roof contours show good quality, which is evidenced by the indexes of completeness and correctness obtained by numerical evaluation. Based on the mean indexes obtained for each experiment, the average completeness and correctness for the experiments were equal to 90.96% and 98.99%, respectively. The maximum completeness and correctness values are 99.19% and 99.94%, respectively, and the minimum values are 78.08% and 97.46%, respectively. The lowest values of completeness are associated to the vegetation occlusion areas and presence of shadows.
FAPESP: 2012/22332-2

This research assesses the accuracy of SRTM and AIRSAR DEMs acquired over the mountainous-hillands of Cameron Highlands with DEMs generated from Digital Aerial Photograph (DAP) with a fine (2 m) spatial resolution and height resolution of about 0.5 m. The ground control points used for generating stereo models from the DAP were acquired during field work using GPS which achieved accuracy better than 2 cm in most cases. To overcome the difficulty of overlaying the DEMs with the DAP DEM as no features can be easily identified on both the images, therefore a technique of using transects and contours generated from the DEMs were used to correct the horizontal displacement. For AIRSAR DEM, comparing the accompanying AIRSAR composite images was also employed. These then allowed an analysis of the height accuracy to be undertaken. The height of both the AIRSAR and SRTM DEMs were also corrected by applying Linear Regression Models. These models were produced by comparing pixels obtained from points, profiles and an area. Once again the corrected DEMs were assessed. Finally the extracted profiles and contours from the corrected SRTM and AIRSAR were compared with the reference DEM. From the comparisons, the horizontal errors were found to be about one and the half pixels (138.72 m: for SRTM) to the east and 1 pixel (10 m: for AIRSAR) to the south. The SD of height differences of the SRTM and AIRSAR DEMs using 90% data were 9.2 m and 5.2 m with profiles comparison; 10.4 m and 5.4 m with area comparison; 10.8 m and 2.4 m with GPS GCPs comparison respectively. From the three comparisons, the means of height differences are 5.2 m, 6.1 m and 15.2 m for SRTM and 8.1 m, 8.3 m and 7.9 m for AIRSAR DEM. The results suggest there is height offset in the AIRSAR DEM. When both heights of DEMs were corrected, the generated contours are close to each other and to reference contours. Using contour colours images and height modelling, the corrected DEM was found to have the potential to detect areas that prone to flash floods and mudslides.

Lately Unmanned Aerial Systems (UAS) are used more frequently in surveying. With broader use comes higher demands on the uncertainty in such measurements. The post processing software is an important factor that affects the uncertainty in the finished product. Therefore it is vital to evaluate how results differentiate in different software and how parameters contribute. In UAS-photogrammetry images are acquired with an overlap which makes it possible to generate point clouds in photogrammetric software. These point clouds are often used to create Digital Terrain Models (DTM).  The purpose of this study is to evaluate how the level of uncertainty differentiates when processing the same UAS-data through block adjustment and dense image matching in two different photogrammetric post processing software. The software used are UAS Master and Pix4D. The objective is also to investigate how the level of extraction in UAS Master and the setting for image scale in Pix4D affects the results when generating point clouds. Three terrain models were created in both software using the same set of data, changing only extraction level and image scale in UAS Master and Pix4D respectively.  26 control profiles were measured with network-RTK in the area of interest to calculate the root mean square (RMS) and mean deviation in order to verify and compare the uncertainty of the terrain models. The study shows that results vary when processing the same UAS-data in different software.  The study also shows that the extraction level in UAS Master and the image scale in Pix4D impacts the results differently. In UAS Master the uncertainty decreases with higher extraction level when generating terrain models. A clear pattern regarding the image scale setting in Pix4D cannot be determined. Both software were able to produce elevation models with a RMS-value of around 0,03 m. The mean deviation in all models created in this study were below 0,02 m, which is the requirement for class 1 in the technical specification SIS-TS 21144:2016. However the mean deviation for the ground type gravel in the terrain model created in UAS Master at a low extraction level exceeds the demands for class 1. This indicates all but one of the created models fulfil the requirements for class 1, which is the class containing the highest requirements.
Obemannade flygfarkostsystem (eng. Unmanned Aerial Systems, UAS) används allt mer frekvent för datainsamling inom geodetisk mätning. I takt med att användningsområdena ökar ställs också högre krav på mätosäkerheten i dessa mätningar. De efterbearbetningsprogram som används är en faktor som påverkar mätosäkerheten i den slutgiltiga produkten. Det är därför viktigt att utvärdera hur olika programvaror påverkar slutresultatet och hur valda parametrar spelar in. I UAS-fotogrammetri tas bilder med övertäckning för att kunna generera punktmoln som i sin tur kan bearbetas till digitala terrängmodeller (DTM).  Syftet med studien är att utvärdera hur mätosäkerheten skiljer sig när samma data bearbetas genom blockutjämning och tät bildmatchning i två olika programvaror. Programvarorna som används i studien är UAS Master och Pix4D. Målet är också att utreda hur vald extraktions nivå i UAS Master och vald bildskala i Pix4D påverkar resultatet vid generering av terrängmodeller. Tre terrängmodeller skapades i UAS Master med olika extraktionsnivåer och ytterligare tre skapades i Pix4D med olika bildskalor. 26 kontrollprofiler mättes in med nätverks-RTK i aktuellt område för beräkning av medelavvikelse och kvadratiskt medelvärde (RMS). Detta för att kunna verifiera och jämföra mätosäkerheten i modellerna. Studien visar att slutresultatet varierar när samma data bearbetas i olika programvaror.  Studien visar också att vald extraktionsnivå i UAS Master och vald bildskala i Pix4D påverkar resultatet olika. I UAS Master minskar mätosäkerheten med ökad extraktionsnivå, i Pix4D är det svårare att se ett tydligt mönster. Båda programvaror kunde producera terrängmodeller med ett RMS-värde kring 0,03 m. Medelavvikelsen i samtliga modeller understiger 0,02 m, vilket är kravet för klass 1 från den tekniska specifikationen SIS-TS 21144:2016. Medelavvikelsen för marktypen grus i UAS Master i modellen med låg extraktionsnivå överskrider dock kraven för klass 1. Därmed uppnår alla förutom en av terrängmodellerna kraven för klass 1, vilket är den klass med högst ställda krav.

Ce travail porte, de façon théorétique et pratique, sur les sujets plus pertinents autour des drones en navigation autonome et semi-autonome. Conformément à la nature multidisciplinaire des problèmes étudies, une grande diversité des techniques et théories ont été couverts dans les domaines de la robotique, l’automatique, l’informatique, la vision par ordinateur et les systèmes embarques, parmi outres.Dans le cadre de cette thèse, deux plates-formes expérimentales ont été développées afin de valider la théorie proposée pour la navigation autonome d’un drone. Le premier prototype, développé au laboratoire, est un quadrirotor spécialement conçu pour les applications extérieures. La deuxième plate-forme est composée d’un quadrirotor à bas coût du type AR.Drone fabrique par Parrot. Le véhicule est connecté sans fil à une station au sol équipé d’un système d’exploitation pour robots (ROS) et dédié à tester, d’une façon facile, rapide et sécurisé, les algorithmes de vision et les stratégies de commande proposés. Les premiers travaux développés ont été basés sur la fusion de donnés pour estimer la position du drone en utilisant des capteurs inertiels et le GPS. Deux stratégies ont été étudiées et appliquées, le Filtre de Kalman Etendu (EKF) et le filtre à Particules (PF). Les deux approches prennent en compte les mesures bruitées de la position de l’UAV, de sa vitesse et de son orientation. On a réalisé une validation numérique pour tester la performance des algorithmes. Une tâche dans le cahier de cette thèse a été de concevoir d’algorithmes de commande pour le suivi de trajectoires ou bien pour la télé-opération. Pour ce faire, on a proposé une loi de commande basée sur l’approche de Mode Glissants à deuxième ordre. Cette technique de commande permet de suivre au quadrirotor de trajectoires désirées et de réaliser l’évitement des collisions frontales si nécessaire. Etant donné que la plate-forme A.R.Drone est équipée d’un auto-pilote d’attitude, nous avons utilisé les angles désirés de roulis et de tangage comme entrées de commande. L’algorithme de commande proposé donne de la robustesse au système en boucle fermée. De plus, une nouvelle technique de vision monoculaire par ordinateur a été utilisée pour la localisation d’un drone. Les informations visuelles sont fusionnées avec les mesures inertielles du drone pour avoir une bonne estimation de sa position. Cette technique utilise l’algorithme PTAM (localisation parallèle et mapping), qui s’agit d’obtenir un nuage de points caractéristiques dans l’image par rapport à une scène qui servira comme repère. Cet algorithme n’utilise pas de cibles, de marqueurs ou de scènes bien définies. La contribution dans cette méthodologie a été de pouvoir utiliser le nuage de points disperse pour détecter possibles obstacles en face du véhicule. Avec cette information nous avons proposé un algorithme de commande pour réaliser l’évitement d’obstacles. Cette loi de commande utilise les champs de potentiel pour calculer une force de répulsion qui sera appliquée au drone. Des expériences en temps réel ont montré la bonne performance du système proposé. Les résultats antérieurs ont motivé la conception et développement d’un drone capable de réaliser en sécurité l’interaction avec les hommes et les suivre de façon autonome. Un classificateur en cascade du type Haar a été utilisé pour détecter le visage d’une personne. Une fois le visage est détecté, on utilise un filtre de Kalman (KF) pour améliorer la détection et un algorithme pour estimer la position relative du visage. Pour réguler la position du drone et la maintenir à une distance désirée du visage, on a utilisé une loi de commande linéaire
The present document addresses, theoretically and experimentally, the most relevant topics for Unmanned Aerial Vehicles (UAVs) in autonomous and semi-autonomous navigation. According with the multidisciplinary nature of the studied problems, a wide range of techniques and theories are covered in the fields of robotics, automatic control, computer science, computer vision and embedded systems, among others. As part of this thesis, two different experimental platforms were developed in order to explore and evaluate various theories and techniques of interest for autonomous navigation. The first prototype is a quadrotor specially designed for outdoor applications and was fully developed in our lab. The second testbed is composed by a non expensive commercial quadrotor kind AR. Drone, wireless connected to a ground station equipped with the Robot Operating System (ROS), and specially intended to test computer vision algorithms and automatic control strategies in an easy, fast and safe way. In addition, this work provides a study of data fusion techniques looking to enhance the UAVs pose estimation provided by commonly used sensors. Two strategies are evaluated in particular, an Extended Kalman Filter (EKF) and a Particle Filter (PF). Both estimators are adapted for the system under consideration, taking into account noisy measurements of the UAV position, velocity and orientation. Simulations show the performance of the developed algorithms while adding noise from real GPS (Global Positioning System) measurements. Safe and accurate navigation for either autonomous trajectory tracking or haptic teleoperation of quadrotors is presented as well. A second order Sliding Mode (2-SM) control algorithm is used to track trajectories while avoiding frontal collisions in autonomous flight. The time-scale separation of the translational and rotational dynamics allows us to design position controllers by giving desired references in the roll and pitch angles, which is suitable for quadrotors equipped with an internal attitude controller. The 2-SM control allows adding robustness to the closed-loop system. A Lyapunov based analysis probes the system stability. Vision algorithms are employed to estimate the pose of the vehicle using only a monocular SLAM (Simultaneous Localization and Mapping) fused with inertial measurements. Distance to potential obstacles is detected and computed using the sparse depth map from the vision algorithm. For teleoperation tests, a haptic device is employed to feedback information to the pilot about possible collisions, by exerting opposite forces. The proposed strategies are successfully tested in real-time experiments, using a low-cost commercial quadrotor. Also, conception and development of a Micro Aerial Vehicle (MAV) able to safely interact with human users by following them autonomously, is achieved in the present work. Once a face is detected by means of a Haar cascade classifier, it is tracked applying a Kalman Filter (KF), and an estimation of the relative position with respect to the face is obtained at a high rate. A linear Proportional Derivative (PD) controller regulates the UAV’s position in order to keep a constant distance to the face, employing as well the extra available information from the embedded UAV’s sensors. Several experiments were carried out through different conditions, showing good performance even under disadvantageous scenarios like outdoor flight, being robust against illumination changes, wind perturbations, image noise and the presence of several faces on the same image. Finally, this thesis deals with the problem of implementing a safe and fast transportation system using an UAV kind quadrotor with a cable suspended load. The objective consists in transporting the load from one place to another, in a fast way and with minimum swing in the cable

[Truncated abstract] Ecologists are increasingly interested in quantifying local interacting processes and their impacts on spatial vegetation patterns. In arid and semiarid ecosystems, theoretical models (often spatially explicit) of dynamical system behaviour have been used to provide insight into changes in vegetation patterning and productivity triggered by ecological events, such as fire and episodic rainfall. The incorporation of aerial imagery of vegetation patterning into current theoretical model remains a challenge, as few theoretical models may be inferred directly from ecological data, let alone imagery. However, if conclusions drawn from theoretical models were well supported by image data then these models could serve as a basis for improved prediction of complex ecosystem behaviour. The objective of this thesis is therefore to innovate methods for inferring theoretical models of vegetation dynamics from imagery. ... These results demonstrate how an ad hoc inference procedure returns biologically meaningful parameter estimates for a germ-grain model of T. triandra vegetation patterning, with VLSA photography as data. Various aspects of the modelling and inference procedures are discussed in the concluding chapter, including possible future extensions and alternative applications for germ-grain models. I conclude that the state-and-transition model provides an effective exploration of an ecosystem?s dynamics, and complements spatially explicit models designed to test specific ecological mechanisms. Significantly, both types of models may now be inferred from image data through the methodologies I have developed, and can provide an empirical basis to theoretical models of complex vegetation dynamics used in understanding and managing arid (and other) ecological systems.

Made available in DSpace on 2017-07-21T14:19:30Z (GMT). No. of bitstreams: 1 Victor Schnepper Lacerda.pdf: 2047594 bytes, checksum: f0234089904caa6e03e22d3efba8394c (MD5) Previous issue date: 2017-02-02
Coordenação de Aperfeiçoamento de Pessoal de Nível Superior
Soybean cultivation is of great importance to the Brazilian economy, and one of the major obstacles to its high productivity is the Asian soybean rust, a disease caused by the fungus Phakopsora pachyrhizi. The main measure to control the damage caused by this disease is the application of fungicides at the appropriate time, but the biggest obstacle to its implementation is the difficult detection of Asian rust in its early stages. In this sense, remote sensing combined with the use of unmanned aerial vehicles (UAVs) has potential for disease detection, especially for providing information that is hard to assess by traditional means, and for the advantages of quality and cost of this technology. The present work explores the use of unmanned aerial vehicles to detect and predict the severity of Asian soybean rust by use of digital image processing and data mining techniques for retrieval of predictive models of severity in different development stages. The models obtained showed satisfactory potential for Asian rust detection, and a high correlation between disease severity and the visible spectrum (RGB camera), as it was possible to obtain correlation coefficients greater than 93% after the R5 development stage of the soybean crop.
O cultivo da soja (Glycine max) é importante para a economia brasileira, sendo que um dos principais obstáculos à alta produtividade na lavoura é a ferrugem asiática, causada pelo fungo Phakopsora pachyrhizi. O principal fator para o controle de danos causados por essa doença é a aplicação de fungicidas em momento apropriado, porém o maior obstáculo para uso dessa medida é a difícil detecção da ferrugem asiática em estágios iniciais. Nesse sentido, o sensoriamento remoto aliado ao uso de veículos aéreos remotamente pilotados apresenta potencial para detecção da doença, principalmente por fornecer informação de difícil acesso aos meios tradicionais e pelas vantagens de qualidade e custo dessa tecnologia. O presente trabalho explora o uso de veículos aéreos remotamente pilotados para detecção e predição de severidade da ferrugem asiática da soja, associados a técnicas de processamento digital de imagens e de mineração de dados, visando a obtenção de modelos preditivos de severidade nos diferentes estágios de desenvolvimento da soja. Os modelos obtidos demonstraram potencial para a detecção da ferrugem asiática, e uma boa correlação da severidade da doença com o espectro visível (câmera RGB), ao passo que foi possível obter coeficientes de correlação maiores que 93% utilizando o algoritmo SMOREG após o estádio R5 de desenvolvimento da cultura da soja.

Made available in DSpace on 2017-07-21T14:19:29Z (GMT). No. of bitstreams: 1 TIAGO GERKE.pdf: 2494737 bytes, checksum: 62b5d2787dcdebe0e9ccf39bfa75f626 (MD5) Previous issue date: 2017-02-03
Wheat cultivation plays an important role to Brazil and the world economic development, as well as in the human diet. The wheat Brazilian production is insufficient to meet the national demand, making research needed in order to improve the yield of this cereal. The goal of this work was to estimate wheat yield, searching for a predictive model through the data mining techniques, with data obtained from high spatial resolution images collected by unmanned aerial vehicles (UAV). The work was carried out in two experimental areas at Ponta Grossa city, Parana state, where for each area eight images were taken, at different culture development stages, with spatial resolution of 3.4cm/px and two images with resolution 10cm/px and 20cm/px, using an eBee UAV with an RGB and a NIR camera. The image processing was done with the Pix4D software, and resulted in an orthomosaics with reflectance values at different wavelengths: Red, Green and Blue, from the RGB camera and Red, Greed and NIR from the NIR camera, besides an image with NDVI values obtained from the arithmetic of NIR and Red wavelengths. The georeferencing correction of each orthomosaic and the extraction of the reflectance values were done with Quantum GIS geographic information system (GIS). From the extracted reflectance values, databases in different proportions (10%, 20%, 40%, 70% and 100%) were created for data mining, using the SMOReg algorithm, based on a support vector machine (SVM) for regression (SVR). The georreferencing correction using 10 control points provided ortomosaics with mean square error (RSME) of distance of 0.35m, which did not show significant difference compared to the correction with 5 control points (RMSE = 0.38m). The reflectance values were different for each study area, making it difficult to indicate better periods for estimating wheat yield. The highest correlation were obtained with data from RGB camera images, followed by the NIR and NDVI camera, with correlations of 0.6168,0.5423 and 0.5324, respectively. The amount of information extracted from the images, reflected in the proportion of the databases, was not significant to generated predictive models, as well as in the correlation indexes, which were statistically the same. Better correlation indices were obtained from the data extracted from the images with spatial resolution of 20cm/px, which suggests that high spatial resolution images may not be adequate for wheat yield estimation.
O cultivo do trigo desempenha um papel importante no desenvolvimento econômico de várias regiões do Brasil e do mundo, bem como na dieta humana. A produção brasileira do trigo é insuficiente para atender à demanda nacional, tornando necessárias pesquisas com intuito de melhorar a produtividade desse cereal. O objetivo desse trabalho foi a estimativa de produtividade do trigo, a partir da criação de modelos preditivos por meio da mineração de dados obtidos em imagens de alta resolução espacial, coletadas por aeronave remotamente pilotada (RPA). O trabalho foi realizado em duas áreas experimentais na cidade de Ponta Grossa – PR, onde para cada área foram feitas oito coletas de imagens, em diferentes estádios de desenvolvimento da cultura, com resolução espacial de 3,4cm/px e duas coletas com resolução 10cm/px e 20cm/px, através de uma RPA eBee utilizando uma câmera RGB e outra NIR. O processamento das imagens foi feito a partir do software Pix4D, e resultou em um ortomosaicos com os valores de refletância em diferentes comprimentos de onda: R, G e B da câmera RGB e R, G e NIR da câmera NIR, além de uma imagem com valores de NDVI obtidos a partir da aritmética das bandas Nir e Red (vermelho). A correção de georreferenciamento de cada ortomosaico e a extração dos valores de refletância foram feitas com auxílio do sistema de informação geográfica (SIG) Quantum GIS. A partir dos valores de refletância extraídos, foram criadas bases de dados em diferentes proporções (10%, 20%, 40%, 70% e 100%) para mineração de dados por meio do algoritmo SMOReg, baseado em máquina de vetor de suporte (SVM) para regressão (SVR). A correção de georreferenciamento utilizando 10 pontos de controle proporcionou ortomosaicos com erro médio quadrático (RSME) de distância de 0,35m, o qual não mostrou diferença significativa para a correção com 5 pontos de controle (RMSE = 0,38m). Os valores de refletância foram diferentes para cada área de estudo, tornando difícil a indicação de melhores períodos para a estimativa de produtividade do trigo. Os maiores índices de correlação da produtividade com os comprimentos de onda, foram obtidos com os dados das imagens da câmera RGB, seguido da câmera NIR e NDVI, com as correlações de 0,6168, 0,5423 e 0,5324, respectivamente. A quantidade de informação extraída das imagens, refletida na proporção das bases de dados, não se mostrou significativa nos modelos preditivos gerados, bem como nos índices de correlação, os quais foram estatisticamente iguais. Índices de correlação melhores foram obtidos a partir dos dados extraídos das imagens com resolução espacial de 20cm/px, o que sugere que imagens de alta resolução espacial podem não ser adequadas para estimativa de produtividade do trigo.

Nesta dissertação é apresentado o desenvolvimento de uma ferramenta computacional para o processamento de pares de imagens estereoscópicas obtidos por câmeras aéreas métricas e não métricas. O programa foi desenvolvido na linguagem C++ e foi utilizado a biblioteca OpenGL. O resultado obtido é uma imagem tridimensional de onde pode ser extraídas cotas de altura e formas de terreno. Estas imagens poderão ser usadas no estudo de áreas de risco em encostas.
In this dissertation is presented the development of a computational tool for the processing of pairs of images estereoscópicas obtained by metric and not metric aerial cameras. The program was developed in the program language C++ and the library was used OpenGL. The result of the program is a three-dimensional image from where it can be extracted height quotas and land forms. These images can be used in the study of risk areas on slopes.

A dissertation submitted to the Faculty of Engineering, University of the Witwatersrand, Johannesburg, in fulfillment of the requirements for the degree of master of Science in Engineering.
A software package was written under the X Window System, Version 11, to assist in manual stereopsis of multiple views. The package enables multiple high resolution (2000 by 1500 pixels and higher) black and white photographs to be viewed simultaneously. Images have adjustable zoom windows which can be manipulated with the pointing device. The zoom windows enlarge to many times the resolution of the image enabling sub-pixel measurements to be extrapolated by the operator. A user-friendly interface allows for fast pinhole camera calibration (from known 3D calibration points) and enables three dimensional lines, circles, grids, cylinders and planes to be fitted to markers specified by the user. These geometric objects are automatically rendered in 3D for comparison with the images. The camera calibration is performed using an iterative optimisation algorithm which also tries multiple combinations of omitted calibration points. This allows for some fault tolerance of the algorithm with respect to erroneous calibration points. Vector mathematics for the geometrical fits is derived. The calibration is shown to converge on a variety of photographs from actual plant surveys. In an artificial test on an array of constructed 3D coordinate markers, absolute accuracy was found to be 1 mm (standard deviation of the Euclidean error) for a distance of 2.5 meters from a standard 35 mm camera. This translates to an error of 1.6 pixels in the scanned views. Lens distortion was assumed to be negligible, except for aspect ratio distortion which was calibrated for. Finally. to demonstrate the efficacy of the package, a 3D model was reconstructed from ten photographs of a human face, taken from different angles.
AC2017

Resource managers frequently require moderate to high resolution imagery within short turnaround periods for use in a GIS-based management system. These spatial data can greatly enhance their ability to make timely, cost-saving decisions and recommendations. MBB Consulting Engineers, Inc., of Pietermaritzburg, South Africa had for many years made use of airborne videography to provide the imagery for several resource-based applications. Applications included detailed land use mapping in various South African river catchments and identification, density classification and mapping of alien vegetation. While the system was low cost and easy to operate, MBB had found that their system was inherently limited, particularly by its lack of automation and poor spatial resolution. This project was started because of a need to address these limitations and provide an airborne remote sensing system that was more automated and could produce higher resolution imagery than the existing system. In addition, the overall cost and time required to produce a map of the resource of interest needed to be reduced. The system developed in this project aimed to improve upon the pre-flight planning and in-flight image acquisition aspects of the existing system. No new post-flight image processing procedures were developed, but possible future refinement of the post-flight image processing routine was considered throughout the development of the system. A pre-flight planning software package was developed that could quickly and efficiently calculate the positions offlight lines and photographs or images with a minimum of user input. The in-flight image acquisition setup developed involved the integration of a high resolution digital still camera, a Global Positioning System (GPS), and camera control software. The use of the rapidly developing and improving technology of a digital still camera was considered to be a better alternative than a video graphic or traditional film camera system for a number of reasons. In particular, digital still cameras produce digital imagery without the need for development and scanning of aerial photographs or frame grabbing of video images. Furthermore, the resolution of current digital still cameras is already significantly better than that of video cameras and is rivalling the resolution of 35rnm film. The system developed was tested by capturing imagery of an urban test area. The images obtained were then rectified using photogrammetric techniques. Results obtained were promising with planimetric accuracies of 5 to 1 Om being obtained. From this test it was concluded that for high accuracy applications involving numerous images, use would be made of softcopy photogrammetric software to semi-automatically position and rectify images, while for applications requiring fewer images and lower accuracy, images could be rectified using the simpler technique of assigning GCPs for each image from scanned orthophotos.
Thesis (MSc.)- University of Natal,Pietermaritzburg, 2000.

碩士
國立成功大學
測量及空間資訊學系
102
Aerial photogrammetry technological development aims to acquire ground measurements rapidly and accurately. Most historical aerial images are difficult to assemble into an image because the pose information of historical images is usually lost. Manual image stitching is essential to solve this problem, but this task is time consuming. Scale invariant feature transform (SIFT) algorithm is widely used to perform feature extraction. This study employed the SIFT algorithm to automate image matching and then used least-squares adjustment to calculate the six parameters of each image on the basis of affine transformation. Through a further study, image exterior parameters were automatically obtained and transformed from the image coordinate system to the ground coordinate system for image registration. We obtained the transformation parameters of each image and the interior accuracy through least-squares adjustment. We then evaluated the external accuracy on the basis of check points. This study successfully used SIFT image matching and least squares for the registration of historical aerial images.

Forest canopies are a critical component of forest ecosystems as they influence many important functions. Specifically, the structure of forest canopies is a driver of the magnitude and rate of these functions. Therefore, being able to accurately measure canopy structure is crucial to ensure ecological models and forest management plans are as robust and efficient as possible. However, canopies are complex and dynamic entities and thus their structure can be challenging to accurately measure. Here we study the feasibility of using lidar to measure forest canopy structure across large spatial extents by investigating the compatibility of aerial and terrestrial lidar systems. Building on known structure-function relationships measured with terrestrial lidar, we establish grounds for scaling these relationships to the aerial scale. This would enable accurate measures of canopy structural complexity to be acquired at landscape and regional scales without the time and labor requirements of terrestrial data collection. Our results illustrate the potential for measures of canopy height, vegetation area, horizontal cover, and canopy roughness to be upscaled. Furthermore, we highlight the benefit of utilizing multivariate measures of canopy structure, and the capacity of lidar to identify forest structural types. Moving forward, lidar is a tool to be utilized in tandem with other technologies to best understand the spatial and temporal dynamics of forests and the influence of physical ecosystem structure.

M.Ing.
This investigation explores the applicability of an adapted formal computational model for rapid synthesis of complete UAV (Unmanned Aerial Vehicle) systems in a single unified environment. The proposed framework termed XPDS (Cross-Platform Data Server) incorporates principles from a variety of similar, successful languages such as Giotto and Esterel. Application of such models has been shown to be advantageous in the UAV control system domain. The proposed solution extends the principles to the complete generic crafts/ground station problem and provides a unified framework for the development of distributed, scalable, and predictable solutions. The core of the framework is a hybrid FLET (Fixed Logical Execution Time) computational model which formalises the timing and operation of a number of concurrent processes or tasks. Three mechanisms are built upon the computational model – a design environment, simulation extensions, and code generation functionality. A design environment is proposed which permits a user to operate through an intuitive interface. The simulation extensions provide tight integration into established software such as Mathwork’s MatLab and Austin Meyer’s X-Plane. The code generation framework allows XPDS programs to be potentially converted into source for a variety of target systems. The combination of the three mechanisms and the formal computational model allow stakeholders to incrementally construct, test, and verify a complete UAV system. An implementation of the proposed framework is constructed to verify the proposed design. Initially, the implementation is subjected to a number of experiments that show that it is a valid representation of the specification. A simplified helicopter stability control system, based upon the problem statement from the initial literature review, is then presented as a test case and the solution is subsequently developed in XPDS. The scenario is successfully constructed and tested through the framework, demonstrating the validity of the proposed solution. The investigation demonstrates that it is both possible and beneficial to develop UAV systems in a single, unified environment. The incorporation of a formal computational model leads to rapid development of predictable solutions. The numerous systems are also easily integrated and benefit from features such as modularity and reusability.

Historically the U.S. Forest Service has used uncorrected aerial photographs to delineate proposed and past management activities on the land base it manages. Transferring a boundary from an image not planimetrically correct to a planimetrically corrected image introduces errors. Positional accuracy of boundaries affects the number of acres the Forest is accountable for managing, and the annual sale quantity (ASQ) or annual board feet targets. The purpose of this study was to develop a methodology that eliminated the need to transfer the boundary from an uncorrected to a corrected image. Raster and vector warping methods were evaluated with reference to positional accuracy and efficiency. Due to the rugged topography of the Siuslaw National Forest, selection of ground control points (GCPs) was an important function in the accurate transformation of images. A Vector warping method, Rubber Sheeting the ARC/INFO projective transformation for all digital GCPs, to all of the Global Position System (GPS) ground control points, provided the most accurate rectification of vector boundaries that had been digitized or scanned from an uncontrolled low elevation photograph.
Graduation date: 1992

M.Phil.
This work proposes a fast local image feature detector and descriptor that is im- plementable on a GPU. The BFROST feature detector is the first published GPU implementation of the popular FAST detector. A simple but novel method of feature orientation estimation which can be calculated in constant time is proposed. The robustness and reliability of the orientation estimation is validated against rotation invariant descriptors such as SIFT and SURF. Furthermore, the BFROST feature descriptor is robust to noise, scalable, rotation invariant, fast to compute in parallel and maintains low memory usage. It is demonstrated that BFROST is usable in real-time applications such as vision-based localization and mapping of images captured from micro aerial platforms.

Indiana University-Purdue University Indianapolis (IUPUI)
Urban forestry is a very important element of urban structures that can improve the environment and life quality within the urban areas. Having an accurate classification of urban forests and grass areas would help improve focused urban tree planting and urban heat wave mitigation efforts. This research project will compare the use of high – resolution aerial imagery and LiDAR data when used to classify canopy and grass areas. The high – resolution image, with 1 – meter resolution, was captured by The National Agriculture Imagery Program (NAIP) on 6/6/2012. Its coordinate system is the North American Datum of 1983 (NAD83). The LiDAR data, with 1.0 – meter average post spacing, was captured by Indiana Statewide Imagery and LiDAR Program from 03/13/2011 to 04/30/2012.The study area is called the Near East Side Community Organization (NESCO) neighborhood. It is located on the east side of downtown Indianapolis, Indiana. Its boundaries are: 65 interstate, East Massachusetts Avenue, East 21st Street, North Emerson Avenue, and the rail road tracks on the south of the East Washington Street. This research will also perform the accuracy assessment based on the results of classifications using high – resolution aerial imagery and LiDAR data in order to determine and explain which method is more accurate to classify urban canopy and grass areas.

Vision-based conventional landing of a fixed wing UAV is addressed in this thesis. The work includes mathematical modeling, interface to a software for rendering the outside scenery, image processing techniques, control law development and outdoor experimentation. This research focuses on detecting the lines or the edges that flank the landing site, use them as visual cues to extract the geometrical parameters such as the line co-ordinates and the line slopes, that are mapped to the control law, to align and conventionally land the fixed wing UAV. Pre-processing and image processing techniques such as Canny Edge detection and Hough Transforms have been used to detect the runway lines or the edges of a landing strip. A Vision-in-the-Loop Simulation (VILS) set up on a personal computer or laptop, has been developed, without any external camera/equipment or networking cables that enables visual serving toper form vision-based studies and simulation. UAV mass, inertia, engine and aero data from literature has been used along withUAV6DOF equations to represent the UAV mathematical model. The UAV model is interfaced to a software using UDP data packets via ports, for rendering the outside scenery in accordance with the UAV’s translation and orientation. The snapshots of the outside scenery, that is passed through an internet URL by including the ‘http’ protocol, is image processed to detect the lines and the line parameters for the control. VILS set has been used to simulate UAV alignment to the runway and landing. Vision-based alignment is achieved by rolling the UAV such that the landing strip that is off center is brought to the center of the image plane. A two stage proportional aileron control input using the line co-ordinates, bringing the midpoints of the top ends of the runway lines to the center of the image, followed by bringing the mid points of the bottom ends of the runway lines to the center of the image has been demonstrated through simulation. A vision-based control for landing has been developed, that consists of an elevator command that is commiserate with the acceptable range of glide slope followed by a flare command till touch down, which is a function of the flare height and estimated height from the 3rd order polynomial of the runway slope obtained by characterization. The feasibility of using the algorithms for a semi-prepared or unprepared landing strip with no visible runway lines have also been demonstrated. Landing on an empty tract of land and in poor visibility condition, by synthetically drawing the runway lines based on a single 3rd order slope. vs height polynomial solution are also presented. A fixed area, and a dynamic area search for the Hough peaks in the Hough accumulator array for the correct detection of lines are addressed. A novel technique for crosswind landing, quite different from conventional techniques, has been introduced, using only the aileron control input for correcting the drift. Three different strategies using the line co-ordinates and the line slopes, with varying levels of accuracy have been presented and compared. About 125 landing data of a manned instrumented prototype aircraft have been analysed to corroborate the findings of this research. Outdoor experiments are also conducted to verify the feasibility of using the line detection algorithm in a realistic scenario and to generate experimental evidence for the findings of this research. Computation time estimates are presented to establish the feasibility of using vision for the problem of conventional landing. The thesis concludes with the findings and direction for future work.