Vectorizing Planar Roof Structure From Very High Resolution Remote Sensing Images Using Transformers

Wufan Zhao; Claudio Persello; Xianwei Lv; Alfred Stein

doi:10.1109/IGARSS52108.2023.10282198

Vectorizing Planar Roof Structure From Very High Resolution Remote Sensing Images Using Transformers

Wufan Zhao^*, Claudio Persello, Xianwei Lv, Alfred Stein

^*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Academic › peer-review

13 Downloads (Pure)

Abstract

Grasping the roof structure of a building is a key part of building reconstruction. Directly predicting the geometric structure of the roof from a raster image to a vectorized representation, however, remains challenging. This paper introduces an efficient and accurate parsing method based upon a vision Transformer we dubbed Roof-Former. Our method consists of three steps: 1) Image encoder and edge node initialization, 2) Image feature fusion with an enhanced segmentation refinement branch, and 3) Edge filtering and structural reasoning. The vertex and edge heat map F1-scores have increased by 2.0% and 1.9% on the VWB dataset when compared to HEAT. Additionally, qualitative evaluations suggest that our method is superior to the current state-of-the-art. It indicates effectiveness for extracting global image information and maintaining the consistency and topological validity of the roof structure.

Original language	English
Title of host publication	IGARSS 2023 - 2023 IEEE International Geoscience and Remote Sensing Symposium, Proceedings
Publisher	IEEE
Pages	4899-4902
Number of pages	4
ISBN (Electronic)	9798350320107
DOIs	https://doi.org/10.1109/IGARSS52108.2023.10282198
Publication status	Published - 2023
Event	43rd IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2023 - Pasadena Convention Center, Pasadena, United States Duration: 16 Jul 2023 → 21 Jul 2023 Conference number: 43 https://2023.ieeeigarss.org/index.php

Publication series

Name	International Geoscience and Remote Sensing Symposium (IGARSS)
Volume	2023-July

Conference

Conference	43rd IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2023
Abbreviated title	IGARSS 2023
Country/Territory	United States
City	Pasadena
Period	16/07/23 → 21/07/23
Internet address	https://2023.ieeeigarss.org/index.php

Keywords

geometry reconstruction
remote sensing image
Roof structure extraction
Transformer
2024 OA procedure

Access to Document

10.1109/IGARSS52108.2023.10282198

Vectorizing_Planar_Roof_Structure_From_Very_High_Resolution_Remote_Sensing_Images_Using_TransformersFinal published version, 10.7 MBLicence: Taverne

Cite this

Zhao, W., Persello, C., Lv, X., & Stein, A. (2023). Vectorizing Planar Roof Structure From Very High Resolution Remote Sensing Images Using Transformers. In IGARSS 2023 - 2023 IEEE International Geoscience and Remote Sensing Symposium, Proceedings (pp. 4899-4902). (International Geoscience and Remote Sensing Symposium (IGARSS); Vol. 2023-July). IEEE. https://doi.org/10.1109/IGARSS52108.2023.10282198

@inproceedings{6dfc7ca52c7b41bdb4c4a47fc5dc41f1,

title = "Vectorizing Planar Roof Structure From Very High Resolution Remote Sensing Images Using Transformers",

abstract = "Grasping the roof structure of a building is a key part of building reconstruction. Directly predicting the geometric structure of the roof from a raster image to a vectorized representation, however, remains challenging. This paper introduces an efficient and accurate parsing method based upon a vision Transformer we dubbed Roof-Former. Our method consists of three steps: 1) Image encoder and edge node initialization, 2) Image feature fusion with an enhanced segmentation refinement branch, and 3) Edge filtering and structural reasoning. The vertex and edge heat map F1-scores have increased by 2.0% and 1.9% on the VWB dataset when compared to HEAT. Additionally, qualitative evaluations suggest that our method is superior to the current state-of-the-art. It indicates effectiveness for extracting global image information and maintaining the consistency and topological validity of the roof structure.",

keywords = "geometry reconstruction, remote sensing image, Roof structure extraction, Transformer, 2024 OA procedure",

author = "Wufan Zhao and Claudio Persello and Xianwei Lv and Alfred Stein",

note = "Publisher Copyright: {\textcopyright} 2023 IEEE.; 43rd IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2023, IGARSS 2023 ; Conference date: 16-07-2023 Through 21-07-2023",

year = "2023",

doi = "10.1109/IGARSS52108.2023.10282198",

language = "English",

series = "International Geoscience and Remote Sensing Symposium (IGARSS)",

publisher = "IEEE",

pages = "4899--4902",

booktitle = "IGARSS 2023 - 2023 IEEE International Geoscience and Remote Sensing Symposium, Proceedings",

address = "United States",

url = "https://2023.ieeeigarss.org/index.php",

}

Zhao, W, Persello, C, Lv, X & Stein, A 2023, Vectorizing Planar Roof Structure From Very High Resolution Remote Sensing Images Using Transformers. in IGARSS 2023 - 2023 IEEE International Geoscience and Remote Sensing Symposium, Proceedings. International Geoscience and Remote Sensing Symposium (IGARSS), vol. 2023-July, IEEE, pp. 4899-4902, 43rd IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2023, Pasadena, California, United States, 16/07/23. https://doi.org/10.1109/IGARSS52108.2023.10282198

Vectorizing Planar Roof Structure From Very High Resolution Remote Sensing Images Using Transformers. / Zhao, Wufan; Persello, Claudio; Lv, Xianwei et al.
IGARSS 2023 - 2023 IEEE International Geoscience and Remote Sensing Symposium, Proceedings. IEEE, 2023. p. 4899-4902 (International Geoscience and Remote Sensing Symposium (IGARSS); Vol. 2023-July).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Academic › peer-review

TY - GEN

T1 - Vectorizing Planar Roof Structure From Very High Resolution Remote Sensing Images Using Transformers

AU - Zhao, Wufan

AU - Persello, Claudio

AU - Lv, Xianwei

AU - Stein, Alfred

N1 - Conference code: 43

PY - 2023

Y1 - 2023

N2 - Grasping the roof structure of a building is a key part of building reconstruction. Directly predicting the geometric structure of the roof from a raster image to a vectorized representation, however, remains challenging. This paper introduces an efficient and accurate parsing method based upon a vision Transformer we dubbed Roof-Former. Our method consists of three steps: 1) Image encoder and edge node initialization, 2) Image feature fusion with an enhanced segmentation refinement branch, and 3) Edge filtering and structural reasoning. The vertex and edge heat map F1-scores have increased by 2.0% and 1.9% on the VWB dataset when compared to HEAT. Additionally, qualitative evaluations suggest that our method is superior to the current state-of-the-art. It indicates effectiveness for extracting global image information and maintaining the consistency and topological validity of the roof structure.

AB - Grasping the roof structure of a building is a key part of building reconstruction. Directly predicting the geometric structure of the roof from a raster image to a vectorized representation, however, remains challenging. This paper introduces an efficient and accurate parsing method based upon a vision Transformer we dubbed Roof-Former. Our method consists of three steps: 1) Image encoder and edge node initialization, 2) Image feature fusion with an enhanced segmentation refinement branch, and 3) Edge filtering and structural reasoning. The vertex and edge heat map F1-scores have increased by 2.0% and 1.9% on the VWB dataset when compared to HEAT. Additionally, qualitative evaluations suggest that our method is superior to the current state-of-the-art. It indicates effectiveness for extracting global image information and maintaining the consistency and topological validity of the roof structure.

KW - geometry reconstruction

KW - remote sensing image

KW - Roof structure extraction

KW - Transformer

KW - 2024 OA procedure

U2 - 10.1109/IGARSS52108.2023.10282198

DO - 10.1109/IGARSS52108.2023.10282198

M3 - Conference contribution

AN - SCOPUS:85178369209

T3 - International Geoscience and Remote Sensing Symposium (IGARSS)

SP - 4899

EP - 4902

BT - IGARSS 2023 - 2023 IEEE International Geoscience and Remote Sensing Symposium, Proceedings

PB - IEEE

T2 - 43rd IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2023

Y2 - 16 July 2023 through 21 July 2023

ER -

Vectorizing Planar Roof Structure From Very High Resolution Remote Sensing Images Using Transformers

Abstract

Publication series

Conference

Keywords

Access to Document

Fingerprint

Cite this