A Hierarchical Building Segmentation in Digital Surface Models for 3D Reconstruction - PubMed Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2017 Jan 24;17(2):222.
doi: 10.3390/s17020222.

A Hierarchical Building Segmentation in Digital Surface Models for 3D Reconstruction

Yiming Yan  1 , Fengjiao Gao  2 , Shupei Deng  3 , Nan Su  4
Affiliations

A Hierarchical Building Segmentation in Digital Surface Models for 3D Reconstruction

Yiming Yan et al. Sensors (Basel). .

Abstract

In this study, a hierarchical method for segmenting buildings in a digital surface model (DSM), which is used in a novel framework for 3D reconstruction, is proposed. Most 3D reconstructions of buildings are model-based. However, the limitations of these methods are overreliance on completeness of the offline-constructed models of buildings, and the completeness is not easily guaranteed since in modern cities buildings can be of a variety of types. Therefore, a model-free framework using high precision DSM and texture-images buildings was introduced. There are two key problems with this framework. The first one is how to accurately extract the buildings from the DSM. Most segmentation methods are limited by either the terrain factors or the difficult choice of parameter-settings. A level-set method are employed to roughly find the building regions in the DSM, and then a recently proposed 'occlusions of random textures model' are used to enhance the local segmentation of the buildings. The second problem is how to generate the facades of buildings. Synergizing with the corresponding texture-images, we propose a roof-contour guided interpolation of building facades. The 3D reconstruction results achieved by airborne-like images and satellites are compared. Experiments show that the segmentation method has good performance, and 3D reconstruction is easily performed by our framework, and better visualization results can be obtained by airborne-like images, which can be further replaced by UAV images.

Keywords: 3D reconstruction; building segmentation; digital surface model; remote sensing.

PubMed Disclaimer

Conflict of interest statement

The authors declare no conflict of interest. The founding sponsors had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, and in the decision to publish the results.

Figures

Figure 1
Figure 1
Flowchart of 3D building reconstruction and visualization.
Figure 2
Figure 2
Examples of four types of buildings. (a) The four buildings; (b) Edge-lines of the roofs.
Figure 3
Figure 3
Blocks of an irregular top building. (a) An irregular top building; (b) Divided blocks.
Figure 4
Figure 4
Height estimation.
Figure 5
Figure 5
Generated building facades with interpolated points.
Figure 6
Figure 6
Hierarchical LS-ORTSEG building segmentation.
Figure 7
Figure 7
Region expansion for level-two segmentation.
Figure 8
Figure 8
Modeling a DSM as an occlusion of textures. (a–e) is one of the five random textures.
Figure 9
Figure 9
Main flow of LS-ORTSEG based EBRs segmentation.
Figure 10
Figure 10
Dataset 1: (a) original DSM shown as image, and 1, 2, 3 and 4 is the serial number of the four buildings; (b) ground truths.
Figure 11
Figure 11
Segmentation results of four buildings by different methods. (a) Level-Set; (b) ORTSEG; (c) LS-ORTSEG; (d) Ground Truths. The ‘Hist.wSize’ is set to 3 for both ORTSEG and LS-ORTSEG.
Figure 12
Figure 12
Accuracy using different window sizes. The ‘NumTextures’ are set to 5 for ORTSEG, and ‘Hist.wSize’ are set from 3 to 9 for both ORTSEG and LS-ORTSEG.
Figure 13
Figure 13
Accuracy using different texture numbers. The ‘Hist.wSize’ are set to 5, and ‘NumTextures’ are set from 2 to 8 for both ORTSEG and LS-ORTSEG.
Figure 14
Figure 14
Original DSM and segmentation results of Dataset 2. (a) DSM (b) Level-set; (c) ORTSEG; (d) LS-ORTSEG. ‘Hist.wSize’ = 3, ‘NumTextures’ = 7 for both ORTSEG and LS-ORTSEG.
Figure 15
Figure 15
Original DSM and segmentation results of Dataset 3. (a) DSM (b) Level-set; (c) ORTSEG; (d) LS-ORTSEG. ‘Hist.wSize’ = 3, ‘NumTextures’ = 7 for both ORTSEG and LS-ORTSEG.
Figure 16
Figure 16
Facade generation of buildings in different cases. (a) Facade generation of Building A by satellite texture-images; (b) facade generation of Building A by airborne-like texture-images; (c) facade generation of Building B by satellite texture-images; (d) façade generation of Building B by airborne-like texture-images.
Figure 17
Figure 17
3D reconstruction of buildings in Dataset 2 and Dataset 3. (a) 3D reconstruction of Building A in multiple views by satellite texture-images; (b) 3D reconstruction of Building A in multiple views by airborne-like texture-images; (c) 3D reconstruction of Building B in multiple views by satellite texture-images; (d) 3D reconstruction of Building B in multiple views by airborne-like texture-images.
Figure 18
Figure 18
Selected points for accuracy verification. (a) Testing points of the two buildings; (b) standard points selected from Google-Earth.
See this image and copyright information in PMC

Similar articles

See all similar articles

Cited by

References

    1. Jin L., Zhu L., Gong H., Pan Y., Du L. Design and development of wetland evaluation system based on remote sensing and 3D visualization; Proceedings of the IEEE International Conference on Computer Application and System Modeling; Taiyuan, China. 22–24 October 2010; pp. 170–173.
    1. Zhang L., Han C., Zhang L., Zhang X., Li J. Web-based visualization of large 3D urban building models. Int. J. Digit. Earth. 2014;7:53–67. doi: 10.1080/17538947.2012.667159. - DOI
    1. Saito K., Spence R. Mapping urban building stocks for vulnerability assessment—Preliminary results. Int. J. Digit. Earth. 2011;4:117–130. doi: 10.1080/17538947.2010.513114. - DOI
    1. Lafarge F., Descombes X., Zerubia J., Pierrot-Deseilligny M. Structural approach for building reconstruction from a single DSM. IEEE Trans. Pattern Anal. Mach. Intell. 2010;32:135–147. doi: 10.1109/TPAMI.2008.281. - DOI - PubMed
    1. Cheng L., Gong J., Li M., Liu Y. 3D building model reconstruction from multi-view aerial imagery and LIDAR data. Photogramm. Eng. Remote Sens. 2011;77:125–139. doi: 10.14358/PERS.77.2.125. - DOI

LinkOut - more resources