A deep learning approach to DTM extraction from imagery using rule-based training labels

C.m. Gevaert (Corresponding Author), C. Persello, F. Nex, G. Vosselman

Research output: Contribution to journalArticleAcademicpeer-review

9 Citations (Scopus)
1 Downloads (Pure)

Abstract


Existing algorithms for Digital Terrain Model (DTM) extraction still face difficulties due to data outliers and geometric ambiguities in the scene such as contiguous off-ground areas or sloped environments. We postulate that in such challenging cases, the radiometric information contained in aerial imagery may be leveraged to distinguish between ground and off-ground objects. We propose a method for DTM extraction from imagery which first applies morphological filters to the Digital Surface Model to obtain candidate ground and off-ground training samples. These samples are used to train a Fully Convolutional Network (FCN) in the second step, which can then be used to identify ground samples for the entire dataset. The proposed method harnesses the power of state-of-the-art deep learning methods, while showing how they can be adapted to the application of DTM extraction by (i) automatically selecting and labelling dataset-specific samples which can be used to train the network, and (ii) adapting the network architecture to consider a larger surface area without unnecessarily increasing the computational burden. The method is successfully tested on four datasets, indicating that the automatic labelling strategy can achieve an accuracy which is comparable to the use of manually labelled training samples. Furthermore, we demonstrate that the proposed method outperforms two reference DTM extraction algorithms in challenging areas.
Original languageEnglish
Pages (from-to)106-123
Number of pages18
JournalISPRS journal of photogrammetry and remote sensing
Volume142
Early online date15 Jun 2018
DOIs
Publication statusPublished - 1 Aug 2018

Fingerprint

digital terrain model
imagery
learning
Labels
education
Labeling
marking
train
aerial photography
harnesses
Antenna grounds
axioms
outlier
Network architecture
ambiguity
surface area
method
Deep learning
Antennas
filter

Keywords

  • ITC-ISI-JOURNAL-ARTICLE

Cite this

@article{8bf9456932bb4338b8575caf69372374,
title = "A deep learning approach to DTM extraction from imagery using rule-based training labels",
abstract = "Existing algorithms for Digital Terrain Model (DTM) extraction still face difficulties due to data outliers and geometric ambiguities in the scene such as contiguous off-ground areas or sloped environments. We postulate that in such challenging cases, the radiometric information contained in aerial imagery may be leveraged to distinguish between ground and off-ground objects. We propose a method for DTM extraction from imagery which first applies morphological filters to the Digital Surface Model to obtain candidate ground and off-ground training samples. These samples are used to train a Fully Convolutional Network (FCN) in the second step, which can then be used to identify ground samples for the entire dataset. The proposed method harnesses the power of state-of-the-art deep learning methods, while showing how they can be adapted to the application of DTM extraction by (i) automatically selecting and labelling dataset-specific samples which can be used to train the network, and (ii) adapting the network architecture to consider a larger surface area without unnecessarily increasing the computational burden. The method is successfully tested on four datasets, indicating that the automatic labelling strategy can achieve an accuracy which is comparable to the use of manually labelled training samples. Furthermore, we demonstrate that the proposed method outperforms two reference DTM extraction algorithms in challenging areas.",
keywords = "ITC-ISI-JOURNAL-ARTICLE",
author = "C.m. Gevaert and C. Persello and F. Nex and G. Vosselman",
year = "2018",
month = "8",
day = "1",
doi = "10.1016/j.isprsjprs.2018.06.001",
language = "English",
volume = "142",
pages = "106--123",
journal = "ISPRS journal of photogrammetry and remote sensing",
issn = "0924-2716",
publisher = "Elsevier",

}

TY - JOUR

T1 - A deep learning approach to DTM extraction from imagery using rule-based training labels

AU - Gevaert, C.m.

AU - Persello, C.

AU - Nex, F.

AU - Vosselman, G.

PY - 2018/8/1

Y1 - 2018/8/1

N2 - Existing algorithms for Digital Terrain Model (DTM) extraction still face difficulties due to data outliers and geometric ambiguities in the scene such as contiguous off-ground areas or sloped environments. We postulate that in such challenging cases, the radiometric information contained in aerial imagery may be leveraged to distinguish between ground and off-ground objects. We propose a method for DTM extraction from imagery which first applies morphological filters to the Digital Surface Model to obtain candidate ground and off-ground training samples. These samples are used to train a Fully Convolutional Network (FCN) in the second step, which can then be used to identify ground samples for the entire dataset. The proposed method harnesses the power of state-of-the-art deep learning methods, while showing how they can be adapted to the application of DTM extraction by (i) automatically selecting and labelling dataset-specific samples which can be used to train the network, and (ii) adapting the network architecture to consider a larger surface area without unnecessarily increasing the computational burden. The method is successfully tested on four datasets, indicating that the automatic labelling strategy can achieve an accuracy which is comparable to the use of manually labelled training samples. Furthermore, we demonstrate that the proposed method outperforms two reference DTM extraction algorithms in challenging areas.

AB - Existing algorithms for Digital Terrain Model (DTM) extraction still face difficulties due to data outliers and geometric ambiguities in the scene such as contiguous off-ground areas or sloped environments. We postulate that in such challenging cases, the radiometric information contained in aerial imagery may be leveraged to distinguish between ground and off-ground objects. We propose a method for DTM extraction from imagery which first applies morphological filters to the Digital Surface Model to obtain candidate ground and off-ground training samples. These samples are used to train a Fully Convolutional Network (FCN) in the second step, which can then be used to identify ground samples for the entire dataset. The proposed method harnesses the power of state-of-the-art deep learning methods, while showing how they can be adapted to the application of DTM extraction by (i) automatically selecting and labelling dataset-specific samples which can be used to train the network, and (ii) adapting the network architecture to consider a larger surface area without unnecessarily increasing the computational burden. The method is successfully tested on four datasets, indicating that the automatic labelling strategy can achieve an accuracy which is comparable to the use of manually labelled training samples. Furthermore, we demonstrate that the proposed method outperforms two reference DTM extraction algorithms in challenging areas.

KW - ITC-ISI-JOURNAL-ARTICLE

UR - https://ezproxy2.utwente.nl/login?url=https://doi.org/http://10.1016/j.isprsjprs.2018.06.001

UR - https://ezproxy2.utwente.nl/login?url=https://webapps.itc.utwente.nl/library/2018/isi/gevaert_dee.pdf

U2 - 10.1016/j.isprsjprs.2018.06.001

DO - 10.1016/j.isprsjprs.2018.06.001

M3 - Article

VL - 142

SP - 106

EP - 123

JO - ISPRS journal of photogrammetry and remote sensing

JF - ISPRS journal of photogrammetry and remote sensing

SN - 0924-2716

ER -