Semantic segmentation in urban scene analysis has mainly focused on images or point clouds, while textured meshes—offering richer spatial representation—remain underexplored. This paper introduces SUM Parts, the first large-scale dataset for urban textured meshes with part-level semantic labels, covering about 2.5 km2with 21 classes. The dataset was created using our own annotation tool, which supports both face- and texture-based annotations with efficient interactive selection. We also provide a comprehensive evaluation of 3D semantic segmentation and interactive annotation methods on this dataset. Our project page is available at https://tudelft3d.github.io/SUMParts/.

Data on the number of floors is required for several applications, for instance, energy demand estimation, population estimation, and flood response plans. Despite this, open data on the number of floors is very rare, even when a 3D city model is available. In practice, it is most often inferred with a geometric method: elevation data is used to estimate the height of a building, which is divided by an assumed storey height and rounded. However, as we demonstrate in this paper with a large dataset of residential buildings, this method is unreliable: <70% of the buildings have a correct estimate. We demonstrate that other attributes and characteristics of buildings can help us better predict the number of floors. We propose several indicators (e.g. construction year, cadastral attributes, building geometry, and neighbourhood census data), and we present a predictive model that was trained with 172,000 buildings in the Netherlands. Our model achieves an accuracy of 94.5% for residential buildings with five floors or less, which is an improvement of about 25% over the geometric approach. Above five floors, our model has only a slight improvement on the geometric approach (5%). The main culprit is the lack of training data for tall buildings, which is uncommon in the Netherlands.

While error propagation in GIS is a topic that has received a lot of attention, it has not been researched with 3D GIS data. We extend error propagation to 3D city models using a Monte Carlo simulation on a use case of annual solar irradiation estimation of building rooftops for assessing the efficiency of installing solar panels. Besides investigating the extension of the theory of error propagation in GIS from 2D to 3D, this paper presents the following contributions. We (1) introduce varying XY/Z accuracy levels of the geometry to reflect actual acquisition outcomes; (2) run experiments on multiple accuracy classes (121 in total); (3) implement an uncertainty engine for simulating acquisition positional errors to procedurally modelled (synthetic) buildings; (4) perform the uncertainty propagation analysis on multiple levels of detail (LODs); and (5) implement Solar3Dcity – a CityGML-compliant software for estimating the solar irradiation of roofs, which we use in our experiments. The results show that in the case of the city of Delft in the Netherlands, a 0.3/0.6 m positional uncertainty yields an error of 68 kWh/m2/year (10%) in solar irradiation estimation. Furthermore, the results indicate that the planar and vertical uncertainties have a different influence on the estimations, and that the results are comparable between LODs. In the experiments we use procedural models, implying that analyses are carried out in a controlled environment where results can be validated. Our uncertainty propagation method and the framework are applicable to other 3D GIS operations and/or use cases. We released Solar3Dcity as open-source software to support related research efforts in the future.

One solution to the integration of additional characteristics, e.g. time and scale, into geographic information system (GIS) datasets is to model them as extra geometric dimensions perpendicular to the spatial ones, creating a higher-dimensional model. Previous work has been mostly limited to higher-dimensional rasters and hierarchies of trees, which grow exponentially with the dimension. As representations with limited topological relationships quickly become intractable in higher dimensions, a topological vector approach seems most suitable for this purpose, requiring the use of higher-dimensional topological data structures. We therefore present in this paper an evaluation and classification of the possible data structures for an nD GIS, including how they can be implemented to support real-world data aspects, such as holes, disconnected components and attributes, as well as practical issues that affect their feasibility, like the availability of algorithms, libraries and software.

One solution to the integration of additional characteristics, for example, time and scale, into GIS data sets is to model them as extra geometric dimensions perpendicular to the spatial ones, creating a higher-dimensional model. While this approach has been previously described and advocated, it is scarcely used in practice because of a lack of high-level construction algorithms and accompanying implementations. We present in this paper a dimension-independent extrusion algorithm permitting us to construct from any (n–1)-dimensional linear cell complex represented as a generalised map, an n-dimensional one by assigning to each (n–1)-cell one or more intervals where it exists along the nth dimension. We have implemented the algorithm in C++11 using CGAL, made the source code publicly available and tested it in experiments using real-world 2D GIS data sets which were extruded to construct up to 5D models.

While there is a growing interest in the use of higher-dimensional (≥4D) digital objects built from complex real-world data, their construction is in practice hampered by a lack of methods and algorithms. We present in this paper a dimension-independent extrusion algorithm for linear geometries using generalised maps. It permits us to create an (n + 1)-dimensional model from an n-dimensional one by assigning it a range along the (n + 1)-th dimension. We present our algorithm, which both optimal and straightforward to implement, using generalised maps as a base, and report on experiments in which 2D real-world GIS datasets were extruded to 3D and 4D.

The knowledge of the transportation mode used by humans (e.g. bicycle, on foot, car and train) is critical for travel behaviour research, transport planning and traffic management. Nowadays, new technologies such as the Global Positioning System have replaced traditional survey methods (paper diaries, telephone) because they are more accurate and problems such as under reporting are avoided. However, although the movement data collected (timestamped positions in digital form) have generally high accuracy, they do not contain the transportation mode. We present in this article a new method for segmenting movement data into single-mode segments and for classifying them according to the transportation mode used. Our fully automatic method differs from previous attempts for five reasons: (1) it relies on fuzzy concepts found in expert systems, that is membership functions and certainty factors; (2) it uses OpenStreetMap data to help the segmentation and classification process; (3) we can distinguish between 10 transportation modes (including between tram, bus and car) and propose a hierarchy; (4) it handles data with signal shortages and noise, and other real-life situations; (5) in our implementation, there is a separation between the reasoning and the knowledge, so that users can easily modify the parameters used and add new transportation modes. We have implemented the method and tested it with a 17-million point data set collected in the Netherlands and elsewhere in Europe. The accuracy of the classification with the developed prototype, determined with the comparison of the classified results with the reference data derived from manual classification, is 91.6%.

One of the simplest methods to construct a 3D city model is to extrude building footprints to obtain ‘block-shaped’ polyhedra representing buildings. Although the method is well known and easy to implement, if the 2D topological relationships between the footprints are not taken into account, the resulting 3D city models will not necessarily be topologically consistent (i.e. primitives shared by 3D buildings will be duplicated and/or intersect each others). As a result, the model will be of little use for most applications, besides visualization that is. In this article, we present a new extrusion procedure to construct topologically consistent 3D city models. It is based on the use of a constrained triangulation, is conceptually simple and offers great flexibility to create city models in different formats (e.g. CityGML or a surface-based representation). We have implemented the procedure, tested it with real-world data sets and validated it.

Fields as found in the geosciences have properties that are not usually found in other disciplines: the phenomena studied are often three‐dimensional (3D), they tend to change continuously over time, and the collection of samples to study the phenomena is problematic, which often results in highly anisotropic distributions of samples. In the geographical information system (GIS) community, raster structures (voxels or octrees) are the most popular solutions, but, as we show in this paper, they have shortcomings for modelling and analysing 3D geoscientific fields. As an alternative to using rasters, we propose a new spatial model based on the Voronoi diagram (VD) and its dual the Delaunay tetrahedralisation (DT), and argue that they have many advantages over other tessellations. We discuss the main properties of the 3D VD/DT, present some GIS operations that are greatly simplified when the VD/DT is used, and, to analyse two or more fields, we also present a variant of the map algebra framework where all the operations are performed directly on VDs. The usefulness of this Voronoi‐based spatial model is demonstrated with a series of potential applications.