Person | TU Delft Repository

R.Y. Peters

View Pure Profile

51 records found

Authored

Building-PCC

Building Point Cloud Completion Benchmarks

Journal article (2024) - W. Gao, R.Y. Peters, J.E. Stoter

With the rapid advancement of 3D sensing technologies, obtaining 3D shape information of objects has become increasingly convenient. Lidar technology, with its capability to accurately capture the 3D information of objects at long distances, has been widely applied in the collect ...

Unsupervised Roofline Extraction from True Orthophotos for LoD2 Building Model Reconstruction

Conference paper (2024) - W. Gao, R.Y. Peters, J.E. Stoter

This paper discusses the reconstruction of LoD2 building models from 2D and 3D data for large-scale urban environments. Traditional methods involve the use of LiDAR point clouds, but due to high costs and long intervals associated with acquiring such data for rapidly developing a ...

面向城市场景异源多时相点云的自动配准

Journal article (2023) - Z. Yang, Z. Yang, Qin Ye, Xufei Wang, R.Y. Peters

Objective Recent advancements in laser scanners and photogrammetry technology have significantly reduced the cost of acquiring 3D point clouds. Consequently, various types of point clouds have gradually become popular data sources for urban applications. The accurate registration of cross-source and multi^-temporal point clouds must be ensured before developing applications based on 3D point clouds. However, this is a challenging task owing to (1) the large amount of data to be considered, (2) the wide discrepancy in characteristics between cross-source point clouds, and (3) the significant changes in a scene represented by multi^-temporal point clouds. These data characteristics can harm the extraction and matching of registration primitives, resulting in the poor performance of marker^-free registration techniques. In this paper, we propose an automated, efficient, and marker^-free method for registering cross-source and multi^-temporal point clouds in urban areas. Methods The proposed registration method comprises three stages: keypoint generation, correspondence matching, and transformation estimation. (1) Keypoint generation. We generate object^-level virtual keypoints as registration primitives rather than directly extracting local features from point clouds, which are redundant and sensitive to outliers and missing data. Specifically, the ground points are first filtered out via the cloth simulation filtering algorithm. The remaining points are decomposed into planar segments by fitting planes in a region-growing manner. Finally, virtual keypoints are determined as the endpoints of intersecting line segments of two adjacent planes. (2) Correspondence matching. First, local triangles are constructed using the generated virtual keypoints as vertices to encode the relative spatial relationships among keypoints within a point cloud. Second, the triangle sets of both point clouds are mapped to a feature space where the triangles become 3D feature points. For each feature point in the source point cloud, we determine its closest point in the target point cloud, forming triangle pairs between the two point clouds. Finally, we propose an improved global matching approach with linear time complexity to extract correspondences encoded in the triangle pairs. (3) Transformation estimation. As cross-source and multi^-temporal point clouds are typically well^-leveled, registration can be achieved by aligning the two point clouds horizontally and translating them vertically. We use the horizontal coordinates of the correspondences to estimate the 2D horizontal transformation and their vertical coordinates to calculate the vertical translation. Results and Discussions We evaluated the effectiveness of the proposed method using large^-scale real^-world urban point clouds. The experimental data consist of six cross-source and multi^-temporal point clouds, including three airborne light detection and ranging (LiDAR) point clouds and three photogrammetric point clouds, which cover an urban area of 1. 8 km² in Rotterdam, the Netherlands. Each point cloud comprises a large number of points (approximately 20‒60 million points per point cloud; refer to Table 1 for details). Additionally, as the point clouds were collected over a long period of time, many of the objects in the scene have changed considerably. These two characteristics make them suitable for performing comprehensive evaluations of automatic marker^-free registration methods. To evaluate the registration results qualitatively, we visualized a randomly selected region (Fig. 7) and three manually selected buildings with varying architectural styles (Fig. 8). Despite the different characteristics of cross-source point clouds and the significant changes in scenes, the proposed method could accurately align all five registration pairs formed by the six experimental point clouds. To evaluate the registration results quantitatively, we calculated both matrix^-based errors (i. e., rotation and translation errors) as well as pointwise errors. The evaluation is summarized in Table 4. Our automatic registration results have an average pointwise error of 6.4 cm, whereas the average matrix^-based errors are 0.2′for rotation and 7.4 cm for translation. Furthermore, despite the massive size of the experimental point clouds, the proposed approach required only 105.7 s to achieve pairwise registration on average. Both qualitative and quantitative results demonstrate the effectiveness of the proposed method for registering cross-source and multi^-temporal urban point clouds. Conclusions A fully automated marker^-free registration approach is presented for cross-source and multi^-temporal point clouds in urban environments. Object^-level virtual keypoints are generated from urban point clouds as registration primitives, thereby overcoming the challenge of identifying valid corresponding features. By encoding rigid body spatial relations among the generated virtual keypoints, we establish correspondences between the source and target point clouds, resulting in efficient matching for large-scale urban scenes. Experiments on real^-world data demonstrate that the proposed method can automatically, accurately, and efficiently register cross-source and multi^-temporal point clouds in urban areas, indicating its practical utility. In the future, we would like to collect more data to test the robustness of the proposed method. Moreover, we intend to study the potential of the proposed matching algorithm in the fusion of general multi^-source data, e. g. , aligning 3D building point clouds with 2D building footprints.

@en

Population Estimation Using a 3D City Model

A Multi-Scale Country-Wide Study in the Netherlands

Journal article (2016) - F. Biljecki, G.A.K. Arroyo Ohori, H. Ledoux, R.Y. Peters, J.E. Stoter

The remote estimation of a region’s population has for decades been a key application of geographic information science in demography. Most studies have used 2D data (maps, satellite imagery) to estimate population avoiding field surveys and questionnaires. As the availability of ...

Contributed

Semantic segmentation of the AHN dataset with the Random Forest Classifier

Master thesis (2021) - M. Papageorgiou, R.Y. Peters, W. Gao

Three-dimensional (3D) city models are of great significance and are high in demand. They can be used for various useful applications such as urban planning, visibility analysis and estimating the solar irradiation and energy demand of buildings throughout the day. Nowadays, Light Detection And Ranging (LiDAR) sensors are one of the most commonly used technologies for the acquisition of cheap, fast, dense, and reliable 3D point cloud datasets. At the same time, increasing attention has been focused lately on the utilization of these 3D point cloud datasets for the reconstruction of 3D city models. The 3D representation of a scene in the form of a 3D point cloud facilitates a variety of analysis tasks like object recognition, segmentation, and classification, which are an important prerequisite for many building reconstruction methods. Thus, having an accurate classifier that can automatically assign 3D points a respective semantic class label is of utmost importance as it can significantly reduce the time and cost required to analyse 3D scenes. Training machine learning and deep learning algorithms to perform this task has been the focus of many recent scientific works that provide promising results and insights for future work.
This thesis attempted to create an accurate Random Forest Classifier for LiDAR point cloud datasets using the Actueel Hoogtebestand Nederland 3 (AHN3) dataset as training data. The aim was to assist building reconstruction methods, and for this reason, only three semantic class labels are assigned to the points by the classifier, namely ground, building and other. Multiple experiments were conducted to test how large the training dataset needs to be, what features should be included and what point density the input point cloud needs to have for the classifier to perform well. Tests were also made using the DALES dataset to evaluate the performance of the classifier for different datasets and environments. Moreover, the time and memory required to train and test the various Random Forests models and the evaluation metrics of the results were stored as benchmarking research to provide insights and guide future work.
The classification approach that was used consists of five major steps. First, the input point cloud is uniformly sampled and then spherical local neighbourhoods of the points are computed in multiple scales. Height and eigen-based features are then extracted for each local neighbourhood, and the best subset of features is then used to train the Random Forest Classifier. Finally, each point of the input point cloud is assigned to the same class as its nearest neighbour in the sampled point cloud. A comparative analysis shows that the performance of our final Random Forest model stands in parallel with that of other available, more complex, deep learning algorithms if we take into consideration its simplicity and efficiency.

Simplification of Massive TINs with the Streaming Geometries Paradigm

Master thesis (2021) - M. de Jong, H. Ledoux, R.Y. Peters

The volume and density of geospatial data is constantly increasing as newer acquisition techniques are developed to create higher-resolution datasets and more countries are creating country-wide datasets such as point clouds. However, methods to deal with this increase in the amount of data are lagging behind and the developments in computer hardware are no longer moving fast enough to compensate. Especially the ability to load and process datasets into the main memory of a computer is becoming more difficult. Therefore, to work with these massive datasets more modern techniques need to be used. In this thesis on one such method for dealing with these massive datasets is explored; streaming simplification of geometries.

Streaming of geometries relies on processing small portions of a geometric dataset at a time, only keeping as much information in memory as is necessary, instead of the whole dataset. This method ensures that memory bottlenecks are prevented and ever-increasingly large datasets can be processed. Processing done on a LIDAR dataset can range from creating rasterized surface or terrain models, to creating a triangulation. Triangulations are an imporant starting point for other analysis techniques such as watershed calculations, line of sight, or noise propagation. However, large datasets of high density lead to large triangulations which are still too complex for these techniques to work with.

To create more manageable triangulations it is necessary to simplify them, which means finding a subset of the vertices that best approximates the original surface. This can be done randomly, or feature-aware by taking into account the shape of the dataset that is being simplified and ensuring data is kept where it is relevant. The AHN3 dataset is so large that simplification cannot be done using traditional techniques. Therefore, to achieve simplification of the AHN3 dataset, this thesis focuses on the modification of an existing methodology for creating streaming triangulations by incorporating a simplification step.

The methodology implemented in this thesis consists of determining where a simplification algorithm can be introduced within the existing streaming pipeline, as well as determining which existing simplification techniques are applicable. After defining possible locations for a simplification module within the streaming pipeline, various methods are created and tested on different sizes of dataset. These methods are: random simplification, decimation, refinement, and a novel medial axis transform based approach. For the refinement method a number of different implementations are investigated.

After the initial results are analyzed, the decimation and medial axis transform algorithms are rejected due to extremely long processing times or a lack of accurate results, respectively. The refinement implementations are able to produce accurate simplified triangulations in a reasonable amount of time, where the best performing method hardly introduces a slowdown in processing at all. Using the fastest algorithm (FCFS) 4 billion points are processed in less than 11 hours, against 13 hours without simplification. Compared to other methods the accuracy of the methodology is similar for all of the approaches. However, the number of points that can be processed per second is less when compared to existing research. The best performing algorithm is a novel method defined as First-Come-First-Serve (FCFS), as it has the best ratio between vertex error and computation time.

The results of this research show that creation and simplification of a massive Delaunay TIN is possible and produces acceptable results, achieving RMSE's below 0.2m consistently. Despite simplification, the size of the resulting datasets is still often too large to fit within the main memory of a computer and will still require streaming processing to perform further calculations on larger areas. This suggests that with the current ever-growing datasets it is relevant to explore how all different types of calculations can be performed in a streaming pipeline, possibly allowing these to be chained one after another.

All methods created for this thesis are found here: \url{https://github.com/mdjong1/simpliPy}

Convolutional Neural Networks for Satellite-Derived Bathymetry

Master thesis (2021) - Y. Yustisi Ardhitasari Lumban Gaol, Ken Arroyo Ohori, R.Y. Peters, A. Amiri Simkooei

Bathymetric depth for shallow water regions is essential for coastal management and research. The measurements using echo sounding and LiDAR leave data gaps because vessels cannot reach nearshore waters or the green laser unable to penetrate specific areas. Satellite-Derived Bathymetry (SDB) is an alternative to extract shallow water depths that is able to overcome these problems using multispectral imagery. There are two approaches of SDB: analytical and empirical. The analytical method requires several water properties, which might not be known. The empirical method relies on the linear relationship between reflectances and depths, but the relationship may not be entirely linear due to bottom types variation, water column, and noise. Machine learning approaches have been used to address nonlinearity, but it treats pixels independently, whereas there is a spatial correlation that influences SDB computations since adjacent pixels are correlated to depth. This characteristic of the local connectivity can be captured by Convolution Neural Networks (CNN). Therefore, this thesis conducts a study of SDB using CNN.

This research focuses on the following questions: (i) what kind of preprocessing is needed for the data sets; (ii) what kind of CNN architecture can be used; (iii) what is the accuracy of the method; and (iv) to what extent the pretrained model in certain areas can be reused in other areas. In order to represent a variety of depth, bottom type, turbidity, and water column properties, this study chooses six areas of interest in three different coastal regions: Puerto Rico, Key West, and Hawaii.

With several CNN configurations, the optimum accuracy is obtained using three convolutional layers, a window size of 9x9, and the RGBNSS bands. Based on the experiment and comparison to the previous studies, the accuracy of SDB using the CNN approach outperforms the linear transform, the ratio transform, Random Forest, and the radiative transfer model. The results show that the accuracy decreases as the depth increases and in more turbid water. Comparison between different image preprocessing indicates another benefit of CNN: removing the need to preprocess images since suitable corrections can be automatically performed by CNN given adequate training.

The use of multi-temporal images enhances the variety of training data and thus improves SDB accuracy. However, data variation should be equally distributed to avoid abnormality in the result. Transfer model analysis indicates several limitations of SDB results at particular depths or when implemented to a different water condition, making the coastal water characteristics considered when reusing a pretrained model from one area to another.

In summary, CNN does not require additional image preprocessing and features specifications for training. CNN can produce better SDB accuracy than several other methods. The accuracy improves by increasing the variety of training data. However, SDB using the transfer model still need to be further investigated. A thorough identification of the proportion of sample data is needed to obtain balanced training data. In this way, it is more likely to produce a more reliable and more stable CNN model for extracting shallow water depths in the new data.

Constructing a digital 3D road network for The Netherlands

Master thesis (2021) - K. Kenesei, R.Y. Peters, J.E. Stoter

This research concerned the design and implementation of a system to perform the 3D conversion of a national road network using airborne Lidar data and land-based road elevation measurements, preserving its topology and quantifying output elevation accuracy. While the system was designed with generality in mind, specific Dutch open data datasets were used as input which were also used in a commercial project with the same goals. In addition to designing a scientifically sound equivalent of the commercial implementation, this research also explored purely academic goals, such as selecting road surface laser reflections accurately and using them for 2.5D road surface modelling purposes. Our system was based on decomposing the 3D conversion task into individually solvable 2.5D sub-problems to enable the use of efficient terrain modelling methods, and to prove that scaling via multiprocessing would be possible. At the 10-30 points per m² sampling density we worked with, the road surfaces are oversampled. We found that because our elevation estimates are based on road reflections only, and because road surfaces are smooth and relatively flat, the output quality and accuracy depends only on the specific technique used to extract the elevations from the Lidar data. Using our methods, the output elevation accuracy represents a 30% increase relative to the Lidar input data, amounting to 10.6 cm at 95% certainty with our specific datasets. This is assumed to be violated only where local density drops below 3 points per m², which occurs where roads are occluded by opaque objects, unless the land-based road surface measurements can provide enough additional data. Elsewhere, more than 98% of the road network conforms with the minimum point density. The 2.5D-based decomposition was found to be an effective means of simplifying the problem and terrain modelling methods were successfully used in the procedure. More than 90% of the traffic-occupied road surfaces are included in the resulting surface models, and about 75% of the total paved surfaces. We extracted the output elevations from them, using linear interpolation where they were unavailable. Our 3D conversion quality and accuracy represents an improvement relative to the commercial results, mostly due to the better prioritising of the input elevation sources, in addition to our solution's capacity to handle complex small to medium-scale occlusion better, including complex relationships between roads such as multi-level motorway junctions.

Semantic segmentation of point clouds with the 3D medial axis transform

Master thesis (2020) - G. Ceccarelli, R.Y. Peters, W. Gao

A point cloud is a representation of shapes, organized in a 3D irregular structure. Point clouds are increasingly used in different applications, ranging from architectural preservation to computer vision. The 3D medial axis transform is a topology preserving, skeleton representation of shapes. It can be used to decompose an object in meaningful parts and to describe local and long range information of points in a point cloud.

In the past years, many deep learning methods for point clouds emerged. These are used for different applications, such as shape classification, object detection or semantic segmentation. In particular, the latter aim to classify each point in the input point cloud in subsets, based on their semantics.

This research investigates the integration of the 3D MAT in two deep learning methods for point clouds' semantic segmentation, PointNet++ and Superpoint Graph. In particular, the 3D MAT was used in PointNet++ as a point feature, to give context to local points. Then, it was used in Superpoint Graph as a geometric descriptor to partition a point cloud and as a edge feature in the SPG.

The major findings of this research outline that the 3D MAT can be successfully used in PointNet++ as a point feature, improving the overall accuracy and loss values of the algorithm. Particularly two MAT derived properties used in this research output positive results, radii and separation angles. These can be combined with point coordinates and RGB information to bring additional knowledge on the geometry of the shape, representing its curvature and thickness. Furthermore, they can be integrated in a simple and effective way, without increasing computational or time effort in the algorithm.

The analysis carried out in Superpoint Graph depicts that the 3D MAT does not improve the initial geometric partition. In fact, adding geometric descriptors to the algorithm increases the difficulty in dividing the point cloud into simple shapes, creating artifacts. Furthermore, adding MAT information on superedges does not give added value to the SPG graph. The reason is that the SPG graph and the structured MAT are different than each other, in practice, as nodes represent diverse parts in the point cloud.

Semantic Segmentation of RGB-Z Aerial Imagery Using Convolutional Neural Networks

Master thesis (2020) - A.E. Mulder, B. Dukai, R.Y. Peters, J.E. Stoter, Sven Briels, Jean Michel Renders

Semantic segmentation (or pixel-level classification) of remotely sensed imagery has shown to be useful for applications in fields as mapping of land cover, object detection, change detection and land-use analysis. Deep learning algorithms called convolutional neural networks (CNNs) have shown to outperform traditional computer vision and machine learning approaches in tackling semantic segmentation tasks. Furthermore, addition of height information (Z) to aerial imagery (RGB) is believed to improve segmentation results. However, discussion remains on the following: to what extent height information adds value; the best way to combine RGB information with height information; and what type of height information can best be used. This study aims to answer these questions. In this research work, the CNN architectures FCN-8s, SegNet, U-Net and FuseNet-SF5 are trained to semantically segment 10 cm resolution true ortho imagery of Haarlem, potentially augmented with height information. The outputted topographic maps contain the classes building, road, water and other. Experiments are conducted that allow for the comparison of 1) models trained on RGB and on RGB-Z, 2) models combining RGB and height information through data fusion and through data stacking, and 3) models trained using different types of absolute and relative height approaches. Performances are compared based on scores on the performance measure (mean) intersection over union (IoU) and through visual assessment of outputted prediction maps. The results indicated that on average segmentation performance improves by approximately 1 percent when absolute height information is added. The class building showed to benefit the most from the addition of height information. Furthermore, extracting features from height information in a separate encoder and fusing these into RGB feature maps, led to a higher overall segmentation quality than when height information is provided as a stacked extra band and processed in the same encoder as the RGB information. Finally, models using relative height delivered a higher quality segmentation than when absolute height approaches were used, especially for large objects. The best performing model; FuseNet-SF5 trained on RGB imagery and pixel-level, relative height, retrieved a mean IoU of 0.8427 and IoUs of 0.8744, 0.7865, 0.9131 and 0.7966 for the classes building, road, water and other respectively. This model was able to correctly classify over 90% of the pixels of 67% of all the objects present in the ground truth. Overall, this study showed that, when considering semantic segmentation of aerial RGB imagery, 1) height information can improve segmentation results, 2) adding height information through data fusion can result in a higher segmentation quality than when data stacking is used, and 3) providing relative height to a network, rather than absolute height, can improve semantic segmentation quality.

Automatic isobath generalisation for navigational charts

Master thesis (2020) - W. van Opstal, B.M. Meijers, R.Y. Peters

Navigational charts play a vital role in a ship's safety while navigating the seas, rivers or lakes. With most of the features and obstructions being out of sight -below sea-level - these charts are more critical than e.g. topographic maps. For routing but also positioning, depth information is a key aspect on these charts. This depth information is available in either depth contours, coloured depth areas and individual soundings. However with the data originating from accurate but usually erratic survey data, a visualisation of raw data is not sufficient for use in a navigational chart directly. It would not clearly convey the information to a human operator in one sight, and thus this visualisation is in need of generalisation: a simplified representation of the same data with irrelevant details being omitted. This thesis gives new insights in the generalisation process for isobaths only and proposes a new framework to deal with those. We propose a framework based on a novel auxiliary data structure to link a triangulation to the resulting isobaths: the triangle region graph. It links the position of isobaths directly to individual triangles, as well as establishes relations between the isobaths themselves. With this structure we ultimately link the survey data to the final cartographic product and thus in theory we could integrate all information across the generalisation pipeline in one and the same process. We have successfully used this framework with a basic rule-based evaluation model: we isolate conflicting isobaths, triangles and vertices based on legibility requirements and target generalisation operators on those. With this approach we can successfully maintain more of the morphology while still yielding a finely legible chart. Especially at large scale charts the results are promising: narrow channels, pits and bends remain if legibility permits. With smaller scale charts the challenge now is to generalise beyond smoothness. More radical generalisation operators are needed to omit all irrelevant details. However the overall framework using the triangle region graph as integrating mechanism has potential to do so. It is easily extensible due to its modular approach and can incorporate most depth information: from survey accuracy to size of isobaths and even golden sounding selection in the future.

Improved AHN3 Gridded DTM/DSM

Student report (2020) - K. Alhoz, K. Kenesei, M. Papageorgiou, E.E.M. Keurentjes, M. de Jong, H. Ledoux, R.Y. Peters

The outcome of the project has been successful from the interpolation perspective, where the final results have shown major improvements over the original rasters, and a multitude of possible other interpolation methods have shown potential. In the end, the algorithm used to crea ...

Automatic detection of waterbeds in shallow muddy water bodies in the Netherlands using green LiDAR

Master thesis (2020) - V. Alexandridis, R.Y. Peters, J.E. Stoter

Bathymetric Airborne LiDAR technology is used to map the depth of water bodies. It uses a green light sensor which is able to penetrate the water surface and reach the bottom part of the interesting water areas. However, water conditions affect the capability of the green laser penetration. Factors such as the water clarity, the water turbidity (waves) and the vegetation are some of the crucial restrictions for green light to penetrate the water; particularly in shallow inland water areas. This research examined the capability of green LiDAR data to improve the bathymetric surveys in case of muddy and shallow inland Dutch water bodies. The potential of green LiDAR increases as the monitoring of water depths is getting easier, faster and more efficiently in terms of cost than manual GPS measurements. The main challenges of this thesis are concentrated both on the existence of various sparse and dense parts in the point-cloud and on the limitations of the data in terms of quality due to the not ideal water conditions. Specifically, this thesis presents a workflow with required procedures that aim to process a raw green LiDAR point clouds of water bodies and then classify them into three classes: water surface, underwater and bottom points. Pulse and Neighbourhood based algorithms were implemented in order to perform a classification process with high level of automation. Point characteristics such as intensity, number of returns, return number were analysed per pulse. Voxelization was used as a spatial method to divide the 3D space into water columns (3D Voxels). The spatial distribution of the water points into the water columns was examined based on different factors such as elevation, density, intensity. By comparing and partially combining those methods the detection process was improved to deal with shallow and muddy water bodies. A classification confidence value was calculated and stored for each potential bottom point. The resulting output is a classified green LiDAR point cloud based on the confidence values. Using elevation, density and confidence values, raster DTMs with multiple bands were created for each water body. To sum up, this thesis proposed an efficient workflow to process and automatically classify green LiDAR water-body data using both voxel and pulse based methods.

Vario-scale visualization of the AHN2 point cloud

Master thesis (2019) - J.D.N. van der Maaden, P.J.M. van Oosterom, B.M. Meijers, H.M.H. van der Heijden, R.Y. Peters

LiDAR technologies are used to measure point cloud data of the earth's surface. The usage of LiDAR allows for the fast collections of massive data sets. The AHN2 point cloud data set, part of Rijkswaterstaats initiative to map the surface of the Netherlands, contains 639 478 217 ...

Visibility Analysis in a Point Cloud based on the Medial Axis Transform

Master thesis (2019) - T. Wu, R.Y. Peters, H. Ledoux

This thesis proposes a novel medial axis transform (MAT) based method to achieve visibility analysis in a point cloud. There are several advantages of this MAT based method. This method avoids surface reconstruction from a point cloud. It also works for the situation when there i ...

Automatic Extraction of an IndoorGML Navigation Graph from an Indoor Point Cloud

Master thesis (2019) - P.L.M. Flikweert, R.Y. Peters, L. Díaz-Vilarino, R.L. Voûte

Because of urbanisation, more than 50% of the global population now lives in cities, a number that is ever-increasing. This leads to a need for more dense and complex building constructions, driven by the fact that people are living increasingly close to one another. These structures are often so large that it becomes difficult to navigate in the environment without any assistance. Especially for people with restricted mobility, such as wheelchair users, or the blind or partially sighted, it can pose a challenge to find the best path in a building. Consequently there is a need for navigation graphs that provide a way to structure connectivity information and routes in an indoor environment. Existing methods generally obtain information on the location of various rooms, corridors and their interconnectivity from 2D floor plans or 3D building models. However, these plans and models often were created by the architect when the building was planned, but never updated with new information after the building was built and came into use. Manually updating them is very labour-intensive. Therefore, in this research a method is developed for obtaining a navigation graph from an indoor environment by automatically extracting the information needed from a point cloud. The navigation graph is modelled according to the Open Geospatial Consortium (OGC) standard IndoorGML, which provides a basic structure for indoor navigation. The point clouds used in this research are gathered with a hand-held Mobile Laser Scanner (MLS), which also saves the path (trajectory) taken in the building. This all leads to the main research question: How can a navigation network in IndoorGML format automatically be extracted from a cluttered indoor point cloud and its trajectory? In order to answer this question this research focuses on the detection of doorways, and how they connect indoor spaces, such as rooms and corridors. A new way of door detection is proposed, which is based on the identification of walls using the 3D Medial Axis Transform (MAT) of the point cloud. After this it is explored how the established connectivity relationships can be extended with accessibility information, such as the location of stairs, and the dimensions of doors. Additionally it is researched how a more detailed navigation graph can be created by subdividing large spaces that have multiple connections. In this thesis it is proven that geometric input for a navigation graph can automatically be obtained from a point cloud. 100% of doors could be detected in the dataset that the methods were developed on, and when testing the methods on another point cloud, this number decreased to 70% , due to unforeseen situations. All spaces connected by detected doors, stairways and sloped surfaces were included in the graph, from which a more extensive navigation graph could be produced by subdividing corridors. This final product was compared to networks drawn by humans on a corresponding floor plan, from which it can be concluded that the graph produced by the methodology of this thesis is a good basis for a navigation.

3D breakline extraction from point clouds with the Medial Axis Transform

Master thesis (2019) - Q. Wang, R.Y. Peters, H. Ledoux

This project develop a new method to generate breakline from the point cloud directly with the MAT. The breakline is a structured line of the object surface which has high curvature. Meanwhile, the reciprocal of the medial edge ball's radius can represent the point with high cur ...

A voxel-based methodology to detect (clustered) outliers in aerial lidar point clouds

Master thesis (2018) - S.J. Griffioen, R.Y. Peters, H. Ledoux, B.M. Meijers

To obtain 3D information of the Earth’s surface, airborne LiDAR technology
is used to quickly capture high-precision measurements of the terrain.
Unfortunately, laser scanning techniques are prone to producing outliers
and noise (i.e. wrong measurements). Therefore, a ...

An image-based method for the pairwise registration of mobile laser scanning point clouds

Master thesis (2018) - A. Christodoulou, P.J.M. van Oosterom, R.Y. Peters, Peter Joosten, Berry van Someren

In this thesis, an image-based method is proposed for solving the relative translation errors of 3D point clouds collected by Mobile Laser Scanning (MLS) techniques. The process of matching 3D point clouds is known as registration. Due to environment-depending limitations of the positioning component of MLS systems, when recording a certain scene more than once, point clouds recorded at different times tend to have mainly positioning distortions. Additionally, due to yaw-angle errors of the recording platform it is possible that the point clouds have small distortions around their Z axis or the point clouds are scanned with fuzziness. This project deals only with the translation errors. Various methods can be found in the literature to perform pair-wise registration of point clouds. Commonly, the challenge of aligning 3D features is tackled in 3D. Only a few techniques for registering point clouds in 2D have being explored. However, the approach presented in this thesis uses the attributes of the 3D points to generate and match 2D-projections, by employing a simple correlation technique instead of matching in 3D. As a result, the developed method depends more on the
number of pixels in the 2D-projections and less on the number of points in the point clouds. This leads to a more cost-efficient method in contrast to 3D registration techniques. The method uses this benefit to provide redundant translation parameters for each point cloud pair. Particularly, several images are created from each point cloud tile. The constructed images illustrate the density of the points, the intensity, the gradient of intensity, the depth, the gradient of the depth and the normal vectors of the points. As a result, the translation parameters are retrieved from the matching of various image-techniques and that is how redundant solutions are provided. Next, with the utilization of image-based evaluation criteria the reliable translation parameters are detected and only those are used to compute the final solutions. The reliable estimations are taken into account for the estimation of the final solutions. Since redundant solutions are provided, the confidence levels of each final estimation can be computed. In addition, an indication of robustness showing how many estimations where included for the computation of the final solution is included. As a result, the developed approach is capable of providing information about the precision and reliability of each pairwise registration. In such a way it is known which of the results can be used in a following step, such as a global registration. Furthermore, a 2D Gaussian elliptical fitting is used to obtain sub-pixel accuracy registration results, as the accuracy of the estimations is restricted to the pixel size of the generated images. It is proven that the developed image-based registration method has the capability to produce reliable matches when there is at least some overlap between two overlapping point clouds and corresponding objects between the point clouds are distinct in pairs of 2D projections. The technique developed for the computation of sub-pixel accuracy results seems to have potential, but further improvement is required.

Bidirectional enrichment of CityGML and Multi-View Stereo Mesh models

Master thesis (2017) - M.S. Tryfona, H. Ledoux, R.Y. Peters

The use of 3D city models has increased the last decades due to the evolution of technology. Their use is related to the need for solutions on issues that correspond to the building environment. Modelling aspects, like geometric, semantic information and topology are necessary, in order to provide an integrated result of spatial analysis. The creation of a complete 3D city model that contains all the needed information for an application, is a time consuming and complex process. Furthermore, different 3D city model formats can contain different aspects of the same features. For example, a CityGML model can have the semantic information of a 3D object, while a Multi-View Stereo Mesh model can contain all the geometric and appearance information of the same 3D object.
This thesis is documenting the possible enrichments that can be done bidirectionally between these two different 3D city model formats, when both exist for the same entities. Moreover, it presents the exploration of possible automated enrichment methodologies that can enrich each 3D city model automatically with information from the other one. Finally, an automatic bidirectional enrichment methodology is proposed and is implemented on a testing area, provided in the two different 3D city model formats. This method is based on distance computations between the meshes of the two 3D city models, used to match corresponding features, or part of features, in order to segment semantically the Multi-View Stereo Mesh model (roof, wall, road, terrain, uncertain) and transfer texture to the surfaces of the CityGML model that correspond with the surfaces of the Multi-View Stereo Mesh model.
In addition, distance computations are performed for the validation of the absence of buildings and the shapes of the roofs in the CityGML model, with respect to the information given from the Multi-View Stereo Mesh model. After the implementation of the proposed methodology, it is found that both 3D city model formats can be used for the proposed enrichments of either 3D city model format.
Future improvements are presented for the achievement of using existing information of different formats of 3D city models of the same entities, in order to supplement each 3D city model format with useful information, enhance the performance of spatial analysis and boost the evolution on the use of such models on real life applications.