The analysis of visibility between two points on the earth's terrain is a common use of GIS software. Most commercial GIS software packages include the ability to generate a viewshed, or a map of terrain surrounding a particular location that would be visible to an observer. Viewsheds are often generated using "bare-earth" Digital Elevation Models (DEMs) derived from the process of photogrammetry. More detailed models, known as Digital Surface Models (DSMs), are often generated using Light Detection and Ranging (LIDAR) which uses an airborne laser to scan the terrain. In addition to having greater accuracy than photogrammetric DEMs, LIDAR DSMs include surface features such as buildings and trees. This project used a visibility algorithm to predict visibility between observer and target locations using both photogrammetric DEMs and LIDAR DSMs of varying resolution. A field survey of the locations was conducted to determine the accuracy of the visibility predictions and to gauge the extent to which the presence of surface features in the DSMs affected the accuracy. The use of different resolution terrain models allowed for the analysis of the relationship between accuracy and optimal grid size. Additionally, a series of visibility predictions were made using Monte Carlo methods to add random error to the terrain elevation to estimate the probability of a target's being visible. Finally, the LIDAR DSMs were used to determine the linear distance of terrain along the lines-of-sight between the observer and targets that were obscured by trees or bushes. A logistic regression was performed between that distance and the visibility of the target to determine the extent to which a greater amount of vegetation along the line-of-sight impacted the target's visibility. / Master of Science

Les appareils d'acquisition d'image 3D sont désormais omniprésents dans plusieurs domaines scientifiques, dont l'imagerie biomédicale, la science des matériaux ou encore l'industrie. La plupart de ces appareils (IRM, scanners à rayons X, micro-tomographes, microscopes confocal, PET scans) produisent un ensemble de données organisées sur une grille régulière que nous nommerons des données digitales, plus couramment des pixels sur des images 2D et des voxels sur des images 3D. Lorsqu'elles sont bien récupérées, ces données approchent la géométrie de la forme capturée (comme des organes en imagerie biomédicale ou des objets dans l'ingénierie). Dans cette thèse, nous nous sommes intéressés à l'extraction de la géométrie sur ces données digitales, et plus précisément, nous nous concentrons à nous approcher des quantités géométriques différentielles comme la courbure sur ces objets. Ces quantités sont les ingrédients critiques de plusieurs applications comme la reconstruction de surface ou la reconnaissance, la correspondance ou la comparaison d'objets. Nous nous focalisons également sur les preuves de convergence asymptotique de ces estimateurs, qui garantissent en quelque sorte la qualité de l'estimation. Plus précisément, lorsque la résolution de l'appareil d'acquisition est augmenté, notre estimation géométrique est plus précise. Notre méthode est basée sur les invariants par intégration et sur l'approximation digitale des intégrations volumiques. Enfin, nous présentons une méthode de classification de la surface, qui analyse les données digitales dans un système à plusieurs échelles et classifie les éléments de surface en trois catégories : les parties lisses, les parties planes, et les parties singulières (discontinuités de la tangente). Ce type de détection de points caractéristiques est utilisé dans plusieurs algorithmes géométriques, comme la compression de maillage ou la reconnaissance d'objet. La stabilité aux paramètres et la robustesse au bruit sont évaluées en fonction des méthodes de la littérature. Tous nos outils pour l'analyse de données digitales sont appliqués à des micro-structures de neige provenant d'un tomographe à rayons X, et leur intérêt est évalué et discuté. / 3D image acquisition devices are now ubiquitous in many domains of science, including biomedical imaging, material science, or manufacturing. Most of these devices (MRI, scanner X, micro-tomography, confocal microscopy, PET scans) produce a set of data organized on a regular grid, which we call digital data, commonly called pixels in 2D images and voxels in 3D images. Properly processed, these data approach the geometry of imaged shapes, like organs in biomedical imagery or objects in engineering. In this thesis, we are interested in extracting the geometry of such digital data, and, more precisely, we focus on approaching geometrical differential quantities such as the curvature of these objects. These quantities are the critical ingredients of several applications like surface reconstruction or object recognition, matching or comparison. We focus on the proof of multigrid convergence of these estimators, which in turn guarantees the quality of estimations. More precisely, when the resolution of the acquisition device is increased, our geometric estimates are more accurate. Our method is based on integral invariants and on digital approximation of volumetric integrals. Finally, we present a surface classification method, which analyzes digital data in a multiscale framework and classifies surface elements into three categories: smooth part, planar part, and singular part (tangent discontinuity). Such feature detection is used in several geometry pipelines, like mesh compression or object recognition. The stability to parameters and the robustness to noise are evaluated with respect to state-of-the-art methods. All our tools for analyzing digital data are applied to 3D X-ray tomography of snow microstructures and their relevance is evaluated and discussed.

M. Ing. / Digital Surface Models (DSM) are widely used in the earth sciences for research, visu- alizations, construction etc. In order to generate a DSM for a speci c area, specialized equipment and personnel are always required which leads to a costly and time consuming exercise. Image processing has become a viable processing technique to generate terrain models since the improvements of hardware provided adequate processing power to complete such a task. Digital Surface Models (DSM) can be generated from stereo imagery, usually obtained from a remote sensing platform. The core component of a DSM generating system is the image matching algorithm. Even though there are a variety of algorithms to date which can generate DSMs, it is a computationally complex calculation and does tend to take some time to complete. In order to achieve faster DSMs, an investigation into an alternative processing platform for the generation of terrain models has been done. The Graphics Processing Unit (GPU) is usually used in the gaming industry to manipulate display data and then render it to a computer screen. The architecture is designed to manipulate large amounts of oating point data. The scientic community has begun using the GPU processing power available for technical computing, hence the term, General Purpose computing on a Graphics Processing Unit (GPGPU). The GPU is investigated as alternative processing platform for the image matching procedure since the processing capability of the GPU is so much higher than the CPU but only for a conditioned set of input data. A matching algorithm, derived from the GC3 algorithm has been implemented on both a CPU platform and a GPU platform in order to investigate the viability of a GPU processing alternative. The algorithm makes use of a Normalized Cross Correlation similarity measurement and the geometry of the image acquisition contained in the sensor model to obtain conjugate point matches in the two source images. The results of the investigation indicated an improvement of up to 70% on the processing time required to generate a DSM. The improvements varied from 70% to some cases where the GPU has taken longer to generate the DSM. The accuracy of the automatic DSM generating system could not be clearly determined since only poor quality reference data was available. It is however shown the DSMs generated using both the CPU and GPU platforms relate to the reference data and correlate to each other. The discrepancies between the CPU and the GPU results are low enough to prove the GPU processing is bene cial with neglible drawbacks in terms of accuracy. The GPU will definitely provide superior processing capabilites for DSM generation above a CPU implementation if a matching algorithm is speci cally designed to cater for the bene ts and limitations of the GPU.

<p> </p> <p>As a result of Shadow NeRF and Sat-NeRF, it is possible to take the solar angle into account in a NeRF-based framework for rendering a scene from a novel viewpoint using satellite images for training. Our work extends those contributions and shows how one can make the renderings season-specific. Our main challenge was creating a Neural Radiance Field (NeRF) that could render seasonal features independently of viewing angle and solar angle</p> <p>while still being able to render shadows. We teach our network to render seasonal features by introducing one more input variable — time of the year. However, the small training datasets typical of satellite imagery can introduce ambiguities in cases where shadows are present in the same location for every image of a particular season. We add additional terms to the loss function to discourage the network from using seasonal features for accounting for shadows. We show the performance of our network on eight Areas of Interest containing images captured by the Maxar WorldView-3 satellite. This evaluation includes tests measuring the ability of our framework to accurately render novel views, generate height maps, predict shadows, and specify seasonal features independently from shadows. Our ablation</p> <p>studies justify the choices made for network design parameters. Also included in this work is a novel approach to space carving which merges multiple features and consistency metrics</p> <p>at different spatial scales to create higher quality digital surface map than is possible using standard RGB features.</p>

No description available.

Flygfotografering med Unmanned Aircraft System (UAS) är i jämförelse med traditionell fotogrammetri effektivare, billigare och säkrare vilket har medfört att denna teknik föredras av många aktörer. Ett tidskrävande arbete som varit svårt att kringgå är att etablera flygsignaler på marken som används för att georeferera och kontrollera flygbilderna med. Under 2018 presenterade UAS-tillverkaren DJI sin nya quadcopter med integrerad Real-Time Kinematic (RTK)-modul. I samband med detta kan kontinuerliga och noggranna positioner levereras via Nätverks-RTK (NRTK) och behovet av markstödpunkter reduceras. I denna studie undersöktes lägesosäkerheterna i plan för ortofoton som framställdes med hjälp av en DJI Phantom 4 RTK där flygbilderna georefererades med begränsat antal eller utan markstödpunkter. Lägesosäkerheterna beräknades och kontrollerades enligt Handbok i mät- och kartfrågor (HMK) – Ortofoto, vilket är ett stöddokument inom ämnet. Vid framställning av ett ortofoto krävs även en digital terrängmodell (DTM) eller en digital ytmodell (Digital Surface Model, DSM) och kvaliteten av denna har stor inverkan på ortofotots kvalitet. I denna studie kontrollerades och utvärderades därför en del av den DSM som användes vid ortofotoframställning för respektive uppsättning enligt den tekniska specifikationen SIS-TS 21144:2016. Resultatet från studien visar att ett ortofoto går att framställas utan markstödpunkter och samtidigt klara kraven på specificerad lägesosäkerhet enligt HMK-standardnivå 3. Den sammanlagda lägesosäkerheten beräknades till 0,029 m vilket är 5 mm högre i jämförelse med ett ortofoto som baserats på traditionell georefereringsmetod, dvs. med markstödpunkter. Kravet på kvalitet i höjddata uppfylldes också för ortofotoframställning trots att en systematisk effekt i höjd uppkom. Denna effekt påverkade inte ortofotots koordinater i plan då standardosäkerheterna i höjd var låga. Resultatet visade att om två markstödpunkter adderades i vardera änden av området, kunde de systematiska effekterna i höjd minimeras och det var då möjligt att skapa en DSM som uppfyller kraven för detaljprojektering (noggrannhetsklass 1–3) enligt SIS-TS 21144:2016. / Aerial photography with UAS is in comparison with traditional photogrammetry more efficient, cheaper and safer which has led to this technology being preferred by many performers. A time-consuming job that has been difficult to avoid is to establish signals at the ground that are used for georeferencing and evaluate the results. In 2018, the UAS manufacturer DJI presented its new quadcopter with integrated Real-Time Kinematic (RTK) module. This allows continuous and accurate positions delivered via Network RTK (NRTK) and the need of ground control points can be reduced. In this study, investigations of the position uncertainties in orthophotos produced using a DJI Phantom 4 RTK carried out where the aerial images were georeferenced with limited numbers or without ground control points. The position uncertainties were calculated and controlled according to the Swedish HMK – Ortofoto (Orthophoto) which is a document within the subject. When producing an orthophoto, a digital terrain model (DTM) or a digital surface model (DSM) is also required and the quality of this has a great impact on the result. Therefore, a part of the DSM used for orthophoto production for each set was checked and evaluated according to the Swedish technical specification, SIS-TS 21144:2016. The result of the study shows that an orthophoto can be produced without ground control points and at the same time meet the requirements for specified position uncertainty according to HMK standard level 3. The total position uncertainty was calculated to be 0,029 m, which is 5 mm higher compared to the orthophoto based on the traditional georeferencing method, i.e. with ground control points. The requirement for quality in height data was also met for orthophoto production even though a systematic effect in height occurred. This effect did not affect the plane coordinates in the orthophoto because of the low standard uncertainties in height. The result showed that if two ground control points were added at each end of the area, the systematic effects were minimized, and it was possible to produce a DSM that fulfils the requirements for accuracy class 1-3 according to SIS-TS 21144:2016.

O sistema de varredura a LASER aerotransportado é uma tecnologia competitiva para levantamentos que visam à geração de modelos digitais de superfície (MDS) e modelos digitais de terreno (MDT). Esta tecnologia sofreu certa resistência por parte das empresas de aerolevantamento e de alguns usuários, porém vem conquistando o mercado, graças à grande quantidade de trabalhos recentemente publicados. Uma das limitações associadas ao levantamento a LASER é a correta detecção de objetos não pertencentes ao terreno como árvores e edificações. Esses objetos, quando devidamente identificados, são essenciais para várias aplicações como cadastro e avaliação da qualidade cartográfica. Este trabalho propõe uma metodologia de identificação de edificações através da detecção de suas bordas, em uma rede triangular (TIN) construída sobre a nuvem de pontos. A utilização da TIN preserva os valores originais da amostra e, representa o universo real com melhor fidelidade que os modelos matriciais. A detecção é realizada calculando-se as declividades para todos os triângulos da TIN. Após esta etapa, são segmentados todos os triângulos que possuem valor de declividade acima de um limiar, pois representam variações abruptas no terreno, condizentes com edificações. A definição do limiar depende da resolução da nuvem de pontos, padrão de varredura e natureza do terreno. O trabalho apresenta ferramentas preliminares para automatizar a vetorização das bordas detectadas. As edificações extraídas pela metodologia e ferramentas apresentadas, podem ser utilizadas para avaliação da qualidade cartográfica do MDS/MDT. Para isso, compara-se o MDS a um documento cartográfico de referência, e realizam-se testes para: detecção e eliminação de erros grosseiros, detecção de tendências e minimização de erros sistemáticos, testes de precisão e de atendimento ao Padrão de Exatidão Cartográfico (PEC). Após análises, conclui-se que a metodologia de detecção de bordas propostas é eficiente, porém as ferramentas de automatização precisam ser aperfeiçoadas. Conclui-se também que o MDS oriundo do levantamento a LASER tem qualidade cartográfica compatível com o produto cartográfico de referência. Recomenda-se aplicação das metodologias propostas em outros conjuntos de dados. / The airborne LASER scanner is competitive technology for surveys that generates digital surface models (DSM) and Digital Terrain Models (DTM). This technology suffered some resistance by the aerosurvey enterprise (and some users), but is conquering its market share due the amount of papers recently published in this field. A limitation associated with this technique is the correct detection of the offterrain objects, like trees and buildings. These objects, when well identified, are essential to a sort of applications like cadastre e map quality evaluation. This dissertation proposes a methodology to identify buildings by the edge detection of a triangular irregular net (TIN) built over the point cloud. The utilization of TIN preserves the original values, representing the real 3D universe with fidelity despite the raster models. The detection is done calculating the slopes for all triangles of the TIN. After that, the triangles with slope values above a threshold are segmented. The Threshold definition depends on resolution of the point cloud, scan pattern and target behavior. Triangles with high slope values represent hard variation and may mean buildings. This work shows preliminary tools to automate the vectorization of the detected edges. The extracted buildings may be used for map quality evaluation of the DSM/DTM, by comparing it with a reference map. This means: test for detection e elimination of bundles, test for detection and minimization of systematic errors, precision tests and evaluate the meeting the Brazilian cartographic accuracy standard (PEC). Theses analysis concludes that the proposed methodology of edge detection is efficient but the developed tools of automation may be sharpened. Also, concludes that the DSM built with LASER scanner survey is compatible with reference map. Its recommended the application of the methodology on other collections of data.

<p>New methods for global shape description of three-dimensional digital objects are presented. The shape of an object is first represented by a digital surface where the faces are either triangles or quadrilaterals. Techniques for computing a high-quality parameterization of the surface are developed and this parameterization is used to approximate the shape of the object. Spherical harmonics are used as basis functions for approximations of the coordinate functions. Information about the global shape is then captured by the coefficients in the spherical harmonics expansions.</p><p>For a starshaped object it is shown how a parameterization can be computed by a projection from its surface onto the unit sphere. An algorithm for computing the position at which the centre of the sphere should be placed, is presented. This algorithm is suited for digital voxel objects. Most of the work is concerned with digital objects whose surfaces are homeomorphic to the sphere. The standard method for computing parameterizations of such surfaces is shown to fail on many objects. This is due to the large distortions of the geometric properties of the surface that often occur with this method. Algorithms to handle this problem are suggested. Non-linear optimization methods are used to find a mapping between a surface and the sphere that minimizes geometric distortion and is useful as a parameterization of the surface. </p><p>The methods can be applied, for example, in medical imaging for shape recognition, detection of shape deformations and shape comparisons of three-dimensional objects.</p>

New methods for global shape description of three-dimensional digital objects are presented. The shape of an object is first represented by a digital surface where the faces are either triangles or quadrilaterals. Techniques for computing a high-quality parameterization of the surface are developed and this parameterization is used to approximate the shape of the object. Spherical harmonics are used as basis functions for approximations of the coordinate functions. Information about the global shape is then captured by the coefficients in the spherical harmonics expansions. For a starshaped object it is shown how a parameterization can be computed by a projection from its surface onto the unit sphere. An algorithm for computing the position at which the centre of the sphere should be placed, is presented. This algorithm is suited for digital voxel objects. Most of the work is concerned with digital objects whose surfaces are homeomorphic to the sphere. The standard method for computing parameterizations of such surfaces is shown to fail on many objects. This is due to the large distortions of the geometric properties of the surface that often occur with this method. Algorithms to handle this problem are suggested. Non-linear optimization methods are used to find a mapping between a surface and the sphere that minimizes geometric distortion and is useful as a parameterization of the surface. The methods can be applied, for example, in medical imaging for shape recognition, detection of shape deformations and shape comparisons of three-dimensional objects.

Generation of Digital Surface Models (DSMs) from stereo satellite (spaceborne) images is classically performed by Ground Control Points (GCPs) which require site visits and precise measurement equipment. However, collection of GCPs is not always possible and such requirement limits the usage of spaceborne imagery. This study aims at developing a fast, fully automatic, GCP-free workflow for DSM generation. The problems caused by GCP-free workflow are overcome using freely-available, low resolution static DSMs (LR-DSM). LR-DSM is registered to the reference satellite image and the registered LR-DSM is used for i) correspondence generation and ii) initial estimate generation for 3-D reconstruction. Novel methods are developed for bias removal for LR-DSM registration and bias equalization for projection functions of satellite imaging. The LR-DSM registration is also shown to be useful for computing the parameters of simple, piecewise empirical projective models. Recent computer vision approaches on stereo correspondence generation and dense depth estimation are tested and adopted for spaceborne DSM generation. The study also presents a complete, fully automatic scheme for GCPfree DSM generation and demonstrates that GCP-free DSM generation is possible and can be performed in much faster time on computers. The resulting DSM can be used in various remote sensing applications including building extraction, disaster monitoring and change detection.