Detection and typification of linear structures for dynamic visualization of 3D city models

https://doi.org/10.1016/j.compenvurbsys.2011.10.001Get rights and content

Abstract

Cluttering is a fundamental problem in 3D city model visualization. In this paper, a novel method for removing cluttering by typification of linear building groups is proposed. This method works in static as well as dynamic visualization of 3D city models. The method starts by converting building models in higher Levels of Details (LoDs) into LoD1 with ground plan and height. Then the Minimum Spanning Tree (MST) is generated according to the distance between the building ground plans. Based on the MST, linear building groups are detected for typification. The typification level of a building group is determined by its distance to the viewpoint as well as its viewing angle. Next, the selected buildings are removed and the remaining ones are adjusted in each group separately. To preserve the building features and their spatial distribution, Attributed Relational Graph (ARG) and Nested Earth Mover’s Distance (NEMD) are used to evaluate the difference between the original building objects and the generalized ones. The experimental results indicate that our method can reduce the number of buildings while preserving the visual similarity of the urban areas.

Introduction

3D city models have been widely employed in numerous applications, and even more increasingly in mobile devices (Rakkolainen and Vainio, 2001, Oulasvirta et al., 2009). Compared with the huge amount of 3D city data, processing resources such as bandwidth and computation ability are relatively limited (Sester and Brenner, 2004, Sun et al., 2009). Therefore, generalization of 3D city models is employed to remove non-essential details and to improve efficiency for storing and processing of these models.

Typification is a generalization operation that replaces a large number of objects by a smaller number while preserving the spatial structure of the objects (Shea & McMaster, 1989). In a city, there are several regular areas suitable for typification, in which the buildings possess a certain common property, e.g., evenly distributed along a road. Therefore, it is critical to detect these regular areas in 3D city models and typify them. The typification in 2D cartographic maps has been studied by many researchers, but less attention has been paid to typification of 3D city models. As indicated by Götzelmann, Guercke, Brenner, and Sester (2009), when changing from 2D to 3D, additional constraints such as terrain and building height should be taken into consideration.

Another main difference from 2D maps is that 3D city models can be viewed from multiple viewpoints, which leads to completely different visualization results. Therefore, the dynamic typification that is required in user interactive 3D city models is proposed in this paper. The relationship between the viewpoint and the 3D city model is mainly considered in the typification process. Arrangements away from the user viewpoint should be typified stronger than those closer to the user viewpoint. Hence, in dynamic environments where the user is allowed to change position, typification has to be carried out dynamically as well. This operation is similar to fisheye lens or focal projections (Hill, 1924), in which the objects far away from the image center are higher abstracted.

The objective of this paper is to detect and typify the linearly structured buildings in 3D city models by taking terrain, building height and different viewpoints into account. A method of evaluating typification results is also proposed based on Attributed Relational Graph (ARG) and Nested Earth Mover’s Distance (NEMD).

The rest of the paper is structured as follows: related work is given in Section 2; Section 3 explains the methodology of the proposed method in detail; experimental results are given in Section 4; Section 5 discusses the proposed methods; and Section 6 concludes the whole paper.

Section snippets

Related work

Automatic cartographic generalization has been acknowledged as important for a long time, since it not only reduces the data volume but also improves the visual presentation of cartographic information in different scales (Mackaness, Ruas, & Sarjakoski, 2007). One of the extensively studied generalization operations is typification of buildings in 2D maps (Anders and Sester, 2000, Neun et al., 2009, Regnauld, 1996, Regnauld, 2001). Various methods were developed for building group typification

Method

The methodology of this research is summarized in Fig. 1 and will be discussed in detail in the following sections: creation of the aligned building groups, typification, and dynamic visualization.

Experimental results

To demonstrate the visualization effects of the proposed dynamic typification algorithm, we test the algorithm on real 3D city models adjusted by artificially generated terrain. The files include simple building models (LoD1) and road lines around Leverkusen in Germany. This section first provides the setup of the experiment system and then presents the visualization results of dynamic typification.

Evaluation

In this section, the typification on single building groups is compared with other methods, and the results of dynamic visualization are discussed in detail.

Conclusions

This paper presents a novel approach to automatic typification of 3D city models using building and terrain features. By extending and dividing the MST of building groups in 3D space, building alignments are effectively detected. These linear structures are then typified by removing the appropriate buildings and adjusting the remaining ones. To preserve the visual similarity, the visual importance of each building in the alignment is evaluated, and the removal order is determined accordingly.

Acknowledgements

We would like to thank the reviewers and the editors for their constructive comments on the paper. This research is part of the ViSuCity Project (PI: Yifang Ban) funded by the Knowledge Foundation, Swedish Foundation for Strategic Research, Vinnova, the Swedish Foundation for Health Care Sciences and Allergy Research, and Invest in Sweden. Bo Mao is supported by a joint fellowship from the Royal Institute of Technology – KTH and the China Scholarship Council and funded by the National Natural

References (38)

  • A. Forberg

    Generalization of 3D building data based on a scale-space approach

    ISPRS Journal of Photogrammetry and Remote Sensing

    (2007)
  • I. Rakkolainen et al.

    A 3D city Info for mobile users

    Computers & Graphics

    (2001)
  • Anders, K. (2005). Level of detail generation of 3D building groups by aggregation and typification. In Proceedings of...
  • K. Anders et al.

    Parameter-free cluster detection in spatial databases and its application to typification

    The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences

    (2000)
  • M. Basaraner et al.

    Structure recognition technique in contextual generalisation of buildings and built-up areas

    The Cartographic Journal

    (2008)
  • D. Burghardt et al.

    Mesh simplification for building typification

    International Journal of Geographical Information Science

    (2007)
  • Christophe, S., & Ruas, A. (2002). Detecting building alignments for generalisation purposes. In International...
  • Fan, H., Meng, L., & Jahnke, M. (2009). Generalization of 3D buildings modeled by CityGML. Lecture notes in...
  • O.D. Faugeras et al.

    Semantic description of aerial images using stochastic labeling

    IEEE Transactions on Pattern Analysis and Machine Intelligence

    (1981)
  • Filippovska, Y., Kada, M., & Fritsch, D. (2009). Quality evaluation of ground plan generalization. In: 12th AGILE...
  • S. Gold et al.

    A graduated assignment algorithm for graph matching

    IEEE Transactions on Pattern Analysis and Machine Intelligence

    (1996)
  • Götzelmann, T., Guercke, R., Brenner, C., & Sester, M. (2009). Terrain-dependent aggregation of 3D city models. In:...
  • Harrie, L. (2001). An optimisation approach to cartographic generalization. Doctor Thesis, Lund...
  • R. Hill

    A lens for whole sky photographs

    Quarterly Journal of the Royal Meteorological Society

    (1924)
  • F.L. Hitchcock

    The distribution of a product from several sources to numerous localities

    Journal of Mathematics and Physics

    (1941)
  • M. Kada

    Automatic generalization of 3D building models

    The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences

    (2002)
  • Kim, D. H., Yun, I. D., & Lee, S. U. (2004). A new attributed relational graph matching algorithm using the nested...
  • T.H. Kolbe

    Representing and exchanging 3D city models with CityGML in 3D geo-information sciences

    (2008)
  • Z. Li et al.

    Automated building generalization based on urban morphology and gestalt theory

    International Journal of Geographical Information Science

    (2004)
  • Cited by (26)

    • An improved LOD specification for 3D building models

      2016, Computers, Environment and Urban Systems
      Citation Excerpt :

      They are usually derived from point clouds or photogrammetry, and their combination with building footprints (Alexander, Smith-Voysey, Jarvis, & Tansey, 2009; Haala & Brenner, 1999; Haala & Kada, 2010). They can also be derived with generalisation from an LOD3 (Mao, Harrie, & Ban, 2012). The models provide a relatively favourable relation between the costs of acquisition and usability.

    • The variants of an LOD of a 3D building model and their influence on spatial analyses

      2016, ISPRS Journal of Photogrammetry and Remote Sensing
      Citation Excerpt :

      We have made a survey of current practices of horizontal and vertical references in LOD1 and LOD2 through an extensive review of research papers that deal with the acquisition of 3D city models, by contacting acquisition companies, and by investigating the specifications of national mapping agencies (NMAs). An overview of research papers such as Mao et al. (2012), He et al. (2012b, 2013), Guercke et al. (2011), Diakité et al. (2014), Commandeur (2012), Oude Elberink et al. (2013), Hermosilla et al. (2012), Schwalbe et al. (2005), Zhang et al. (2006), Brasebin et al. (2016) yielded an extensive overview of the references, but also strengthened our impression that the majority of research papers that describe methods to acquire buildings do not explicitly elaborate on the employed geometric reference. The specifications of local governments and national mapping agencies contained some information about the geometric references.

    • Semantic volume texture for virtual city building model visualisation

      2015, Computers, Environment and Urban Systems
      Citation Excerpt :

      The increased demand for residential space promotes a more flexible illustrative visualisation for inspecting, managing and planning purposes on the refined virtual building models. As a result, there are more demands on interactive illustration or visual exploration of the indoor environments (Hong et al., 2015; Isikdag, Zlatanova, & Underwood, 2013; Goetz, 2012) and three-dimensional (3D) cadastre (van Oosterom, 2013) models, rather than static manual illustration artworks (Zhou, Dao, Thill, & Delmelle, 2015; Shojaei, Kalantari, Bishop, Rajabifard, & Aien, 2013; Wang, Pouliot, & Hubert, 2012; Mao, Harrie, & Ban, 2012; Zhu et al., 2011). However, direct illustrative exploration as cross sections or cutaways of the layered building environment is confronted with some fundamental challenges.

    • Semi-automated approach to indoor mapping for 3D as-built building information modeling

      2015, Computers, Environment and Urban Systems
      Citation Excerpt :

      Finally, in the concluding Section 5, the strengths and limitations of the proposed method are discussed. 3D models have been widely employed in numerous urban system domain applications, such as urban planning, disaster management, navigation systems, and the mobile service industry (Mao, Harrie, & Ban, 2012; Over, Schilling, Neubauer, & Zipf, 2010). 3D representation can enhance understanding of complex spatial and functional relationships in the contemporary 3D city; thus, the demand for high-complexity 3D urban modeling has increased, as driven by recent industry developments.

    • Generalization of 3D building texture using image compression and multiple representation data structure

      2013, ISPRS Journal of Photogrammetry and Remote Sensing
      Citation Excerpt :

      He then extended the simplification method to the roof structures using pre-defined roof types (Kada, 2007). Mao et al. (2012) generalized linear building groups based on visual similarity. Fan and Meng (2012) proposed a three-step approach to generalize 3D buildings.

    View all citing articles on Scopus
    View full text