Computer vision is a research field in which images are the main object of study. One of its category of problems is shape description. Object classification is one important example of applications using shape descriptors. Usually, these processes were performed on 2D images. With the large-scale development of new technologies and the affordable price of equipment that generates 3D images, computer vision has adapted to this new scenario, expanding the classic 2D methods to 3D. However, it is important to highlight that 2D methods are mostly dependent on the variation of illumination and color, while 3D sensors provide depth, structure/3D shape and topological information beyond color. Thus, different methods of shape descriptors and robust attributes extraction were studied, from which new attribute extraction methods have been proposed and described based on 3D data. The results obtained from well known public datasets have demonstrated their efficiency and that they compete with other state-of-the-art methods in this area: the RPHSD (a method proposed in this dissertation), achieved 85:4% of accuracy on the University of Washington RGB-D dataset, being the second best accuracy on this dataset; the COMSD (another proposed method) has achieved 82:3% of accuracy, standing at the seventh position in the rank; and the CNSD (another proposed method) at the ninth position. Also, the RPHSD and COMSD methods have relatively small processing complexity, so they achieve high accuracy with low computing time. / Visão computacional é uma área de pesquisa em que as imagens são o principal objeto de estudo. Um dos problemas abordados é o da descrição de formatos (em inglês, shapes). Classificação de objetos é um importante exemplo de aplicação que usa descritores de shapes. Classicamente, esses processos eram realizados em imagens 2D. Com o desenvolvimento em larga escala de novas tecnologias e o barateamento dos equipamentos que geram imagens 3D, a visão computacional se adaptou para este novo cenário, expandindo os métodos 2D clássicos para 3D. Entretanto, estes métodos são, majoritariamente, dependentes da variação de iluminação e de cor, enquanto os sensores 3D fornecem informações de profundidade, shape 3D e topologia, além da cor. Assim, foram estudados diferentes métodos de classificação de objetos e extração de atributos robustos, onde a partir destes são propostos e descritos novos métodos de extração de atributos a partir de dados 3D. Os resultados obtidos utilizando bases de dados 3D públicas conhecidas demonstraram a eficiência dos métodos propóstos e que os mesmos competem com outros métodos no estado-da-arte: o RPHSD (um dos métodos propostos) atingiu 85:4% de acurácia, sendo a segunda maior acurácia neste banco de dados; o COMSD (outro método proposto) atingiu 82:3% de acurácia, se posicionando na sétima posição do ranking; e o CNSD (outro método proposto) em nono lugar. Além disso, os métodos RPHSD têm uma complexidade de processamento relativamente baixa. Assim, eles atingem uma alta acurácia com um pequeno tempo de processamento.

Computer vision is a research field in which images are the main object of study. One of its category of problems is shape description. Object classification is one important example of applications using shape descriptors. Usually, these processes were performed on 2D images. With the large-scale development of new technologies and the affordable price of equipment that generates 3D images, computer vision has adapted to this new scenario, expanding the classic 2D methods to 3D. However, it is important to highlight that 2D methods are mostly dependent on the variation of illumination and color, while 3D sensors provide depth, structure/3D shape and topological information beyond color. Thus, different methods of shape descriptors and robust attributes extraction were studied, from which new attribute extraction methods have been proposed and described based on 3D data. The results obtained from well known public datasets have demonstrated their efficiency and that they compete with other state-of-the-art methods in this area: the RPHSD (a method proposed in this dissertation), achieved 85:4% of accuracy on the University of Washington RGB-D dataset, being the second best accuracy on this dataset; the COMSD (another proposed method) has achieved 82:3% of accuracy, standing at the seventh position in the rank; and the CNSD (another proposed method) at the ninth position. Also, the RPHSD and COMSD methods have relatively small processing complexity, so they achieve high accuracy with low computing time. / Visão computacional é uma área de pesquisa em que as imagens são o principal objeto de estudo. Um dos problemas abordados é o da descrição de formatos (em inglês, shapes). Classificação de objetos é um importante exemplo de aplicação que usa descritores de shapes. Classicamente, esses processos eram realizados em imagens 2D. Com o desenvolvimento em larga escala de novas tecnologias e o barateamento dos equipamentos que geram imagens 3D, a visão computacional se adaptou para este novo cenário, expandindo os métodos 2D clássicos para 3D. Entretanto, estes métodos são, majoritariamente, dependentes da variação de iluminação e de cor, enquanto os sensores 3D fornecem informações de profundidade, shape 3D e topologia, além da cor. Assim, foram estudados diferentes métodos de classificação de objetos e extração de atributos robustos, onde a partir destes são propostos e descritos novos métodos de extração de atributos a partir de dados 3D. Os resultados obtidos utilizando bases de dados 3D públicas conhecidas demonstraram a eficiência dos métodos propóstos e que os mesmos competem com outros métodos no estado-da-arte: o RPHSD (um dos métodos propostos) atingiu 85:4% de acurácia, sendo a segunda maior acurácia neste banco de dados; o COMSD (outro método proposto) atingiu 82:3% de acurácia, se posicionando na sétima posição do ranking; e o CNSD (outro método proposto) em nono lugar. Além disso, os métodos RPHSD têm uma complexidade de processamento relativamente baixa. Assim, eles atingem uma alta acurácia com um pequeno tempo de processamento.

Background. Domain adaptation is described as, a model learning from a source data distribution and performing well on the target data. This concept, Domain adaptation is applied to assembly-line production tasks to perform an automatic quality inspection. Objectives. The aim of this master thesis is to apply this concept of 3D domain adaptation from synthetic images to real images. It is an attempt to bridge the gap between different domains (synthetic and real point cloud images), by implementing deep learning models that learn from synthetic 3D point cloud (CAD model images) and perform well on the actual 3D point cloud (3D Camera images). Methods. Through this course of thesis project, various methods for understand- ing the data and analyzing it for bridging the gap between CAD and CAM to make them similar is looked into. Literature review and controlled experiment are research methodologies followed during implementation. In this project, we experiment with four different deep learning models with data generated and compare their performance to know which deep learning model performs best for the data. Results. The results are explained through metrics i.e, accuracy and train time, which were the outcomes of each of the deep learning models after the experiment. These metrics are illustrated in the form of graphs for comparative analysis between the models on which the data is trained and tested on. PointDAN showed better results with higher accuracy compared to the other 3 models. Conclusions. The results attained show that domain adaptation for synthetic images to real images is possible with the data generated. PointDAN deep learning model which focuses on local feature alignment and global feature alignment with single-view point data shows better results with our data.

Yes / This paper proposes a stable and accurate object registration pipeline for markerless augmented reality applications. We present two novel algorithms for object recognition and matching to improve the registration accuracy from model to scene transformation via point cloud fusion. Whilst the first algorithm effectively deals with simple scenes with few object occlusions, the second algorithm handles cluttered scenes with partial occlusions for robust real-time object recognition and matching. The computational framework includes a locally supported Gaussian weight function to enable repeatable detection of 3D descriptors. We apply a bilateral filtering and outlier removal to preserve edges of point cloud and remove some interference points in order to increase matching accuracy. Extensive experiments have been carried to compare the proposed algorithms with four most used methods. Results show improved performance of the algorithms in terms of computational speed, camera tracking and object matching errors in semi-cluttered and partial-occluded scenes. / Shanxi Natural Science and Technology Foundation of China, grant number 2016JZ026 and grant number 2016KW-043).

Research on forest ecosystems receives high attention, especially nowadays with regard to sustainable management of renewable resources and the climate change. In particular, accurate information on the 3D structure of a tree is important for forest science and bioclimatology, but also in the scope of commercial applications. Conventional methods to measure geometric plant features are labor- and time-intensive. For detailed analysis, trees have to be cut down, which is often undesirable. Here, Terrestrial Laser Scanning (TLS) provides a particularly attractive tool because of its contactless measurement technique. The object geometry is reproduced as a 3D point cloud. The objective of this thesis is the automatic retrieval of the spatial structure of trees from TLS data. We focus on forest scenes with comparably high stand density and with many occlusions resulting from it. The varying level of detail of TLS data poses a big challenge. We present two fully automatic methods to obtain skeletal structures from scanned trees that have complementary properties. First, we explain a method that retrieves the entire tree skeleton from 3D data of co-registered scans. The branching structure is obtained from a voxel space representation by searching paths from branch tips to the trunk. The trunk is determined in advance from the 3D points. The skeleton of a tree is generated as a 3D line graph. Besides 3D coordinates and range, a scan provides 2D indices from the intensity image for each measurement. This is exploited in the second method that processes individual scans. Furthermore, we introduce a novel concept to manage TLS data that facilitated the researchwork. Initially, the range image is segmented into connected components. We describe a procedure to retrieve the boundary of a component that is capable of tracing inner depth discontinuities. A 2D skeleton is generated from the boundary information and used to decompose the component into sub components. A Principal Curve is computed from the 3D point set that is associated with a sub component. The skeletal structure of a connected component is summarized as a set of polylines. Objective evaluation of the results remains an open problem because the task itself is ill-defined: There exists no clear definition of what the true skeleton should be w.r.t. a given point set. Consequently, we are not able to assess the correctness of the methods quantitatively, but have to rely on visual assessment of results and provide a thorough discussion of the particularities of both methods. We present experiment results of both methods. The first method efficiently retrieves full skeletons of trees, which approximate the branching structure. The level of detail is mainly governed by the voxel space and therefore, smaller branches are reproduced inadequately. The second method retrieves partial skeletons of a tree with high reproduction accuracy. The method is sensitive to noise in the boundary, but the results are very promising. There are plenty of possibilities to enhance the method’s robustness. The combination of the strengths of both presented methods needs to be investigated further and may lead to a robust way to obtain complete tree skeletons from TLS data automatically. / Die Erforschung des ÖkosystemsWald spielt gerade heutzutage im Hinblick auf den nachhaltigen Umgang mit nachwachsenden Rohstoffen und den Klimawandel eine große Rolle. Insbesondere die exakte Beschreibung der dreidimensionalen Struktur eines Baumes ist wichtig für die Forstwissenschaften und Bioklimatologie, aber auch im Rahmen kommerzieller Anwendungen. Die konventionellen Methoden um geometrische Pflanzenmerkmale zu messen sind arbeitsintensiv und zeitaufwändig. Für eine genaue Analyse müssen Bäume gefällt werden, was oft unerwünscht ist. Hierbei bietet sich das Terrestrische Laserscanning (TLS) als besonders attraktives Werkzeug aufgrund seines kontaktlosen Messprinzips an. Die Objektgeometrie wird als 3D-Punktwolke wiedergegeben. Basierend darauf ist das Ziel der Arbeit die automatische Bestimmung der räumlichen Baumstruktur aus TLS-Daten. Der Fokus liegt dabei auf Waldszenen mit vergleichsweise hoher Bestandesdichte und mit zahlreichen daraus resultierenden Verdeckungen. Die Auswertung dieser TLS-Daten, die einen unterschiedlichen Grad an Detailreichtum aufweisen, stellt eine große Herausforderung dar. Zwei vollautomatische Methoden zur Generierung von Skelettstrukturen von gescannten Bäumen, welche komplementäre Eigenschaften besitzen, werden vorgestellt. Bei der ersten Methode wird das Gesamtskelett eines Baumes aus 3D-Daten von registrierten Scans bestimmt. Die Aststruktur wird von einer Voxelraum-Repräsentation abgeleitet indem Pfade von Astspitzen zum Stamm gesucht werden. Der Stamm wird im Voraus aus den 3D-Punkten rekonstruiert. Das Baumskelett wird als 3D-Liniengraph erzeugt. Für jeden gemessenen Punkt stellt ein Scan neben 3D-Koordinaten und Distanzwerten auch 2D-Indizes zur Verfügung, die sich aus dem Intensitätsbild ergeben. Bei der zweiten Methode, die auf Einzelscans arbeitet, wird dies ausgenutzt. Außerdem wird ein neuartiges Konzept zum Management von TLS-Daten beschrieben, welches die Forschungsarbeit erleichtert hat. Zunächst wird das Tiefenbild in Komponenten aufgeteilt. Es wird eine Prozedur zur Bestimmung von Komponentenkonturen vorgestellt, die in der Lage ist innere Tiefendiskontinuitäten zu verfolgen. Von der Konturinformation wird ein 2D-Skelett generiert, welches benutzt wird um die Komponente in Teilkomponenten zu zerlegen. Von der 3D-Punktmenge, die mit einer Teilkomponente assoziiert ist, wird eine Principal Curve berechnet. Die Skelettstruktur einer Komponente im Tiefenbild wird als Menge von Polylinien zusammengefasst. Die objektive Evaluation der Resultate stellt weiterhin ein ungelöstes Problem dar, weil die Aufgabe selbst nicht klar erfassbar ist: Es existiert keine eindeutige Definition davon was das wahre Skelett in Bezug auf eine gegebene Punktmenge sein sollte. Die Korrektheit der Methoden kann daher nicht quantitativ beschrieben werden. Aus diesem Grund, können die Ergebnisse nur visuell beurteiltwerden. Weiterhinwerden die Charakteristiken beider Methoden eingehend diskutiert. Es werden Experimentresultate beider Methoden vorgestellt. Die erste Methode bestimmt effizient das Skelett eines Baumes, welches die Aststruktur approximiert. Der Detaillierungsgrad wird hauptsächlich durch den Voxelraum bestimmt, weshalb kleinere Äste nicht angemessen reproduziert werden. Die zweite Methode rekonstruiert Teilskelette eines Baums mit hoher Detailtreue. Die Methode reagiert sensibel auf Rauschen in der Kontur, dennoch sind die Ergebnisse vielversprechend. Es gibt eine Vielzahl von Möglichkeiten die Robustheit der Methode zu verbessern. Die Kombination der Stärken von beiden präsentierten Methoden sollte weiter untersucht werden und kann zu einem robusteren Ansatz führen um vollständige Baumskelette automatisch aus TLS-Daten zu generieren.

Im Rahmen einer Diplomarbeit an der TU Bergakademie Freiberg wurden in 2014 die Grundlagen für die Auswertung von 3D-Punktwolken zur automatisierten Nachtragung des Risswerks gelegt. Um die dafür notwendigen 3D-Punktwolken möglichst wirtschaftlich zu erstellen, laufen seit 2015 Untersuchungen und Testmessungen zur Machbarkeit des untertägigen Einsatzes von mobil messenden Laserscannern. Im Folgenden werden verschiedene technische Ansätze sowie die Ergebnisse der Testmessungen und die weiteren geplanten Schritte vorgestellt. / As part of a thesis at the Technical University of Freiberg, a basis for the analysis of 3D point clouds was set for refining the mine map automatically. Since 2015 studies and test measurements have been running to create the necessary 3D point clouds as economically as possible, by using an underground mobile scanning system. Below the different technical approaches will be presented as well as the results of the test measurements and the next planned steps.

Im Rahmen einer Diplomarbeit an der TU Bergakademie Freiberg wurden in 2014 die Grundlagen für die Auswertung von 3D-Punktwolken zur automatisierten Nachtragung des Risswerks gelegt. Um die dafür notwendigen 3D-Punktwolken möglichst wirtschaftlich zu erstellen, laufen seit 2015 Untersuchungen und Testmessungen zur Machbarkeit des untertägigen Einsatzes von mobil messenden Laserscannern. Im Folgenden werden verschiedene technische Ansätze sowie die Ergebnisse der Testmessungen und die weiteren geplanten Schritte vorgestellt. / As part of a thesis at the Technical University of Freiberg, a basis for the analysis of 3D point clouds was set for refining the mine map automatically. Since 2015 studies and test measurements have been running to create the necessary 3D point clouds as economically as possible, by using an underground mobile scanning system. Below the different technical approaches will be presented as well as the results of the test measurements and the next planned steps.

Yes / Recovering high-quality 3D point clouds from monocular endoscopic images is a challenging task. This paper proposes a novel deep learning-based computational framework for 3D point cloud reconstruction from single monocular endoscopic images. An unsupervised mono-depth learning network is used to generate depth information from monocular images. Given a single mono endoscopic image, the network is capable of depicting a depth map. The depth map is then used to recover a dense 3D point cloud. A generative Endo-AE network based on an auto-encoder is trained to repair defects of the dense point cloud by generating the best representation from the incomplete data. The performance of the proposed framework is evaluated against state-of-the-art learning-based methods. The results are also compared with non-learning based stereo 3D reconstruction algorithms. Our proposed methods outperform both the state-of-the-art learning-based and non-learning based methods for 3D point cloud reconstruction. The Endo-AE model for point cloud completion can generate high-quality, dense 3D endoscopic point clouds from incomplete point clouds with holes. Our framework is able to recover complete 3D point clouds with the missing rate of information up to 60%. Five large medical in-vivo databases of 3D point clouds of real endoscopic scenes have been generated and two synthetic 3D medical datasets are created. We have made these datasets publicly available for researchers free of charge. The proposed computational framework can produce high-quality and dense 3D point clouds from single mono-endoscopy images for augmented reality, virtual reality and other computer-mediated medical applications.

Research on forest ecosystems receives high attention, especially nowadays with regard to sustainable management of renewable resources and the climate change. In particular, accurate information on the 3D structure of a tree is important for forest science and bioclimatology, but also in the scope of commercial applications. Conventional methods to measure geometric plant features are labor- and time-intensive. For detailed analysis, trees have to be cut down, which is often undesirable. Here, Terrestrial Laser Scanning (TLS) provides a particularly attractive tool because of its contactless measurement technique. The object geometry is reproduced as a 3D point cloud. The objective of this thesis is the automatic retrieval of the spatial structure of trees from TLS data. We focus on forest scenes with comparably high stand density and with many occlusions resulting from it. The varying level of detail of TLS data poses a big challenge. We present two fully automatic methods to obtain skeletal structures from scanned trees that have complementary properties. First, we explain a method that retrieves the entire tree skeleton from 3D data of co-registered scans. The branching structure is obtained from a voxel space representation by searching paths from branch tips to the trunk. The trunk is determined in advance from the 3D points. The skeleton of a tree is generated as a 3D line graph. Besides 3D coordinates and range, a scan provides 2D indices from the intensity image for each measurement. This is exploited in the second method that processes individual scans. Furthermore, we introduce a novel concept to manage TLS data that facilitated the researchwork. Initially, the range image is segmented into connected components. We describe a procedure to retrieve the boundary of a component that is capable of tracing inner depth discontinuities. A 2D skeleton is generated from the boundary information and used to decompose the component into sub components. A Principal Curve is computed from the 3D point set that is associated with a sub component. The skeletal structure of a connected component is summarized as a set of polylines. Objective evaluation of the results remains an open problem because the task itself is ill-defined: There exists no clear definition of what the true skeleton should be w.r.t. a given point set. Consequently, we are not able to assess the correctness of the methods quantitatively, but have to rely on visual assessment of results and provide a thorough discussion of the particularities of both methods. We present experiment results of both methods. The first method efficiently retrieves full skeletons of trees, which approximate the branching structure. The level of detail is mainly governed by the voxel space and therefore, smaller branches are reproduced inadequately. The second method retrieves partial skeletons of a tree with high reproduction accuracy. The method is sensitive to noise in the boundary, but the results are very promising. There are plenty of possibilities to enhance the method’s robustness. The combination of the strengths of both presented methods needs to be investigated further and may lead to a robust way to obtain complete tree skeletons from TLS data automatically. / Die Erforschung des ÖkosystemsWald spielt gerade heutzutage im Hinblick auf den nachhaltigen Umgang mit nachwachsenden Rohstoffen und den Klimawandel eine große Rolle. Insbesondere die exakte Beschreibung der dreidimensionalen Struktur eines Baumes ist wichtig für die Forstwissenschaften und Bioklimatologie, aber auch im Rahmen kommerzieller Anwendungen. Die konventionellen Methoden um geometrische Pflanzenmerkmale zu messen sind arbeitsintensiv und zeitaufwändig. Für eine genaue Analyse müssen Bäume gefällt werden, was oft unerwünscht ist. Hierbei bietet sich das Terrestrische Laserscanning (TLS) als besonders attraktives Werkzeug aufgrund seines kontaktlosen Messprinzips an. Die Objektgeometrie wird als 3D-Punktwolke wiedergegeben. Basierend darauf ist das Ziel der Arbeit die automatische Bestimmung der räumlichen Baumstruktur aus TLS-Daten. Der Fokus liegt dabei auf Waldszenen mit vergleichsweise hoher Bestandesdichte und mit zahlreichen daraus resultierenden Verdeckungen. Die Auswertung dieser TLS-Daten, die einen unterschiedlichen Grad an Detailreichtum aufweisen, stellt eine große Herausforderung dar. Zwei vollautomatische Methoden zur Generierung von Skelettstrukturen von gescannten Bäumen, welche komplementäre Eigenschaften besitzen, werden vorgestellt. Bei der ersten Methode wird das Gesamtskelett eines Baumes aus 3D-Daten von registrierten Scans bestimmt. Die Aststruktur wird von einer Voxelraum-Repräsentation abgeleitet indem Pfade von Astspitzen zum Stamm gesucht werden. Der Stamm wird im Voraus aus den 3D-Punkten rekonstruiert. Das Baumskelett wird als 3D-Liniengraph erzeugt. Für jeden gemessenen Punkt stellt ein Scan neben 3D-Koordinaten und Distanzwerten auch 2D-Indizes zur Verfügung, die sich aus dem Intensitätsbild ergeben. Bei der zweiten Methode, die auf Einzelscans arbeitet, wird dies ausgenutzt. Außerdem wird ein neuartiges Konzept zum Management von TLS-Daten beschrieben, welches die Forschungsarbeit erleichtert hat. Zunächst wird das Tiefenbild in Komponenten aufgeteilt. Es wird eine Prozedur zur Bestimmung von Komponentenkonturen vorgestellt, die in der Lage ist innere Tiefendiskontinuitäten zu verfolgen. Von der Konturinformation wird ein 2D-Skelett generiert, welches benutzt wird um die Komponente in Teilkomponenten zu zerlegen. Von der 3D-Punktmenge, die mit einer Teilkomponente assoziiert ist, wird eine Principal Curve berechnet. Die Skelettstruktur einer Komponente im Tiefenbild wird als Menge von Polylinien zusammengefasst. Die objektive Evaluation der Resultate stellt weiterhin ein ungelöstes Problem dar, weil die Aufgabe selbst nicht klar erfassbar ist: Es existiert keine eindeutige Definition davon was das wahre Skelett in Bezug auf eine gegebene Punktmenge sein sollte. Die Korrektheit der Methoden kann daher nicht quantitativ beschrieben werden. Aus diesem Grund, können die Ergebnisse nur visuell beurteiltwerden. Weiterhinwerden die Charakteristiken beider Methoden eingehend diskutiert. Es werden Experimentresultate beider Methoden vorgestellt. Die erste Methode bestimmt effizient das Skelett eines Baumes, welches die Aststruktur approximiert. Der Detaillierungsgrad wird hauptsächlich durch den Voxelraum bestimmt, weshalb kleinere Äste nicht angemessen reproduziert werden. Die zweite Methode rekonstruiert Teilskelette eines Baums mit hoher Detailtreue. Die Methode reagiert sensibel auf Rauschen in der Kontur, dennoch sind die Ergebnisse vielversprechend. Es gibt eine Vielzahl von Möglichkeiten die Robustheit der Methode zu verbessern. Die Kombination der Stärken von beiden präsentierten Methoden sollte weiter untersucht werden und kann zu einem robusteren Ansatz führen um vollständige Baumskelette automatisch aus TLS-Daten zu generieren.

A large number of current mobile robots use 3D sensors as part of their sensor setup. Common 3D sensors, i.e., laser scanners or RGB-D cameras, emit a signal (laser light or infrared light for instance), and its reflection is recorded in order to estimate depth to a surface. The resulting set of measurement points is commonly referred to as 'point clouds'. In the first part of this dissertation an inherent problem of sensors that emit some light signal is addressed, namely that these signals can be reflected and/or refracted by transparent of highly specular surfaces, causing erroneous or missing measurements. A novel heuristic approach is introduced how such objects may nevertheless be identified and their size and shape reconstructed by fusing information from several viewpoints of the scene. In contrast to other existing approaches no prior knowledge about the objects is required nor is the shape of the reconstructed objects restricted to a limited set of geometric primitives. The thesis proceeds to illustrate problems caused by sensor noise and registration errors and introduces mechanisms to address these problems. Finally a quantitative comparison between equivalent directly measured objects, the reconstructions and "ground truth" is provided. The second part of the thesis addresses the problem of automatically determining the quality of the registration for a pair of point clouds. Although a different topic, these two problems are closely related, if modeled in the fashion of this thesis. After illustrating why the output parameters of a popular registration algorithm (ICP) are not suitable to deduce registration quality, several heuristic measures are developed that provide better insight. Experiments performed on different datasets were performed to showcase the applicability of the proposed measures in different scenarios.