Change detection between digital surface models from airborne laser scanning and dense matching using convolutional neural networks

Z. Zhang, G. Vosselman, M. Gerke, C. Persello, D. Tuia, M.Y. Yang

Research output: Chapter in Book/Report/Conference proceedingConference contributionAcademicpeer-review

Abstract

Airborne photogrammetry and airborne laser scanning are two commonly used technologies used for topographical data acquisition at the city level. Change detection between airborne laser scanning data and photogrammetric data is challenging since the two point clouds show different characteristics. After comparing the two types of point clouds, this paper proposes a feed-forward Convolutional Neural Network (CNN) to detect building changes between them. The motivation from an application point of view is that the multimodal point clouds might be available for different epochs. Our method contains three steps: First, the point clouds and orthoimages are converted to raster images. Second, square patches are cropped from raster images and then fed into CNN for change detection. Finally, the original change map is post-processed with a simple connected component analysis. Experimental results show that the patch-based recall rate reaches 0.8146 and the precision rate reaches 0.7632. Object-based evaluation shows that 74 out of 86 building changes are correctly detected.

LanguageEnglish
Title of host publicationISPRS Geospatial Week 2019, 10–14 June 2019, Enschede, The Netherlands
PublisherISPRS
Pages453-460
Number of pages8
DOIs
Publication statusPublished - 29 May 2019
Event4th ISPRS Geospatial Week 2019 - Enschede, Netherlands
Duration: 10 Jun 201914 Jun 2019

Publication series

NameISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences
VolumeIV-2/W5
ISSN (Print)2194-9042

Conference

Conference4th ISPRS Geospatial Week 2019
CountryNetherlands
CityEnschede
Period10/06/1914/06/19

Fingerprint

airborne lasers
change detection
raster
laser
Neural networks
Scanning
scanning
Photogrammetry
Lasers
photogrammetry
data acquisition
Data acquisition
time measurement
rate
detection
evaluation
lasers
analysis
city
method

Keywords

  • Airborne Laser Scanning
  • Change Detection
  • Convolutional Neural Network (CNN)
  • Dense Image Matching
  • Digital Surface Model (DSM)

Cite this

Zhang, Z., Vosselman, G., Gerke, M., Persello, C., Tuia, D., & Yang, M. Y. (2019). Change detection between digital surface models from airborne laser scanning and dense matching using convolutional neural networks. In ISPRS Geospatial Week 2019, 10–14 June 2019, Enschede, The Netherlands (pp. 453-460). (ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences; Vol. IV-2/W5). ISPRS. https://doi.org/10.5194/isprs-annals-IV-2-W5-453-2019
Zhang, Z. ; Vosselman, G. ; Gerke, M. ; Persello, C. ; Tuia, D. ; Yang, M.Y. / Change detection between digital surface models from airborne laser scanning and dense matching using convolutional neural networks. ISPRS Geospatial Week 2019, 10–14 June 2019, Enschede, The Netherlands. ISPRS, 2019. pp. 453-460 (ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences).
@inproceedings{18f32e9ab3a4443fbd9542a91df9a558,
title = "Change detection between digital surface models from airborne laser scanning and dense matching using convolutional neural networks",
abstract = "Airborne photogrammetry and airborne laser scanning are two commonly used technologies used for topographical data acquisition at the city level. Change detection between airborne laser scanning data and photogrammetric data is challenging since the two point clouds show different characteristics. After comparing the two types of point clouds, this paper proposes a feed-forward Convolutional Neural Network (CNN) to detect building changes between them. The motivation from an application point of view is that the multimodal point clouds might be available for different epochs. Our method contains three steps: First, the point clouds and orthoimages are converted to raster images. Second, square patches are cropped from raster images and then fed into CNN for change detection. Finally, the original change map is post-processed with a simple connected component analysis. Experimental results show that the patch-based recall rate reaches 0.8146 and the precision rate reaches 0.7632. Object-based evaluation shows that 74 out of 86 building changes are correctly detected.",
keywords = "Airborne Laser Scanning, Change Detection, Convolutional Neural Network (CNN), Dense Image Matching, Digital Surface Model (DSM)",
author = "Z. Zhang and G. Vosselman and M. Gerke and C. Persello and D. Tuia and M.Y. Yang",
year = "2019",
month = "5",
day = "29",
doi = "10.5194/isprs-annals-IV-2-W5-453-2019",
language = "English",
series = "ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences",
publisher = "ISPRS",
pages = "453--460",
booktitle = "ISPRS Geospatial Week 2019, 10–14 June 2019, Enschede, The Netherlands",

}

Zhang, Z, Vosselman, G, Gerke, M, Persello, C, Tuia, D & Yang, MY 2019, Change detection between digital surface models from airborne laser scanning and dense matching using convolutional neural networks. in ISPRS Geospatial Week 2019, 10–14 June 2019, Enschede, The Netherlands. ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences, vol. IV-2/W5, ISPRS, pp. 453-460, 4th ISPRS Geospatial Week 2019, Enschede, Netherlands, 10/06/19. https://doi.org/10.5194/isprs-annals-IV-2-W5-453-2019

Change detection between digital surface models from airborne laser scanning and dense matching using convolutional neural networks. / Zhang, Z.; Vosselman, G.; Gerke, M.; Persello, C.; Tuia, D.; Yang, M.Y.

ISPRS Geospatial Week 2019, 10–14 June 2019, Enschede, The Netherlands. ISPRS, 2019. p. 453-460 (ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences; Vol. IV-2/W5).

Research output: Chapter in Book/Report/Conference proceedingConference contributionAcademicpeer-review

TY - GEN

T1 - Change detection between digital surface models from airborne laser scanning and dense matching using convolutional neural networks

AU - Zhang, Z.

AU - Vosselman, G.

AU - Gerke, M.

AU - Persello, C.

AU - Tuia, D.

AU - Yang, M.Y.

PY - 2019/5/29

Y1 - 2019/5/29

N2 - Airborne photogrammetry and airborne laser scanning are two commonly used technologies used for topographical data acquisition at the city level. Change detection between airborne laser scanning data and photogrammetric data is challenging since the two point clouds show different characteristics. After comparing the two types of point clouds, this paper proposes a feed-forward Convolutional Neural Network (CNN) to detect building changes between them. The motivation from an application point of view is that the multimodal point clouds might be available for different epochs. Our method contains three steps: First, the point clouds and orthoimages are converted to raster images. Second, square patches are cropped from raster images and then fed into CNN for change detection. Finally, the original change map is post-processed with a simple connected component analysis. Experimental results show that the patch-based recall rate reaches 0.8146 and the precision rate reaches 0.7632. Object-based evaluation shows that 74 out of 86 building changes are correctly detected.

AB - Airborne photogrammetry and airborne laser scanning are two commonly used technologies used for topographical data acquisition at the city level. Change detection between airborne laser scanning data and photogrammetric data is challenging since the two point clouds show different characteristics. After comparing the two types of point clouds, this paper proposes a feed-forward Convolutional Neural Network (CNN) to detect building changes between them. The motivation from an application point of view is that the multimodal point clouds might be available for different epochs. Our method contains three steps: First, the point clouds and orthoimages are converted to raster images. Second, square patches are cropped from raster images and then fed into CNN for change detection. Finally, the original change map is post-processed with a simple connected component analysis. Experimental results show that the patch-based recall rate reaches 0.8146 and the precision rate reaches 0.7632. Object-based evaluation shows that 74 out of 86 building changes are correctly detected.

KW - Airborne Laser Scanning

KW - Change Detection

KW - Convolutional Neural Network (CNN)

KW - Dense Image Matching

KW - Digital Surface Model (DSM)

U2 - 10.5194/isprs-annals-IV-2-W5-453-2019

DO - 10.5194/isprs-annals-IV-2-W5-453-2019

M3 - Conference contribution

T3 - ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences

SP - 453

EP - 460

BT - ISPRS Geospatial Week 2019, 10–14 June 2019, Enschede, The Netherlands

PB - ISPRS

ER -

Zhang Z, Vosselman G, Gerke M, Persello C, Tuia D, Yang MY. Change detection between digital surface models from airborne laser scanning and dense matching using convolutional neural networks. In ISPRS Geospatial Week 2019, 10–14 June 2019, Enschede, The Netherlands. ISPRS. 2019. p. 453-460. (ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences). https://doi.org/10.5194/isprs-annals-IV-2-W5-453-2019