Details
| Originalsprache | Englisch |
|---|---|
| Seiten (von - bis) | 221-229 |
| Seitenumfang | 9 |
| Fachzeitschrift | ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences |
| Jahrgang | 10 |
| Ausgabenummer | G-2025 |
| Publikationsstatus | Veröffentlicht - 10 Juli 2025 |
| Veranstaltung | 2025 International Society for Photogrammetry and Remote Sensing (ISPRS) Geospatial Week, GSW 2025 - Dubai, Vereinigte Arabische Emirate Dauer: 6 Apr. 2025 → 11 Apr. 2025 |
Abstract
Knowledge about past and present land cover is of interest for the assessment of the current status of our environment and, thus, for proper planning of the future. Information on past land cover is exclusively contained in an implicit way in historic remote sensing imagery and historic topographic maps. To make this information explicit, pixel-wise classification methods based on neural networks can be used. The method proposed in this paper aims to automatically predict land cover based on historic aerial imagery and scanned topographic maps. The proposed deep learning-based classifier extracts features at different scales from both modalities and fuses the most complex topographic map features of the smallest scale to enrich the ones derived from the aerial images. Both, the multi-modal features and those of the aerial images at larger scales, are mapped to pixel-wise predictions by means of a decoder. Comprehensive experiments show that the result of the proposed multi-modal classifier are superior compared to those of a uni-modal aerial image classifier; the multi-modal mIOU of 82.3% is 1.4% larger than the one of uni-modal classifier. This demonstrates that aerial image classification can benefit from additional information contained in topographic maps.
ASJC Scopus Sachgebiete
- Physik und Astronomie (insg.)
- Instrumentierung
- Umweltwissenschaften (insg.)
- Umweltwissenschaften (sonstige)
- Erdkunde und Planetologie (insg.)
- Erdkunde und Planetologie (sonstige)
Zitieren
- Standard
- Harvard
- Apa
- Vancouver
- BibTex
- RIS
in: ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Jahrgang 10, Nr. G-2025, 10.07.2025, S. 221-229.
Publikation: Beitrag in Fachzeitschrift › Konferenzaufsatz in Fachzeitschrift › Forschung › Peer-Review
}
TY - JOUR
T1 - Multi-modal Land Cover Classification of Historical Aerial Images and Topographic Maps Exploiting Attention-based Feature Fusion
AU - Dorozynski, Mareike
AU - Rottensteiner, Franz
AU - Thiemann, Frank
AU - Sester, Monika
AU - Dahms, Thorsten
AU - Hovenbitzer, Michael
N1 - Publisher Copyright: Copyright © 2025 Mareike Dorozynski et al.
PY - 2025/7/10
Y1 - 2025/7/10
N2 - Knowledge about past and present land cover is of interest for the assessment of the current status of our environment and, thus, for proper planning of the future. Information on past land cover is exclusively contained in an implicit way in historic remote sensing imagery and historic topographic maps. To make this information explicit, pixel-wise classification methods based on neural networks can be used. The method proposed in this paper aims to automatically predict land cover based on historic aerial imagery and scanned topographic maps. The proposed deep learning-based classifier extracts features at different scales from both modalities and fuses the most complex topographic map features of the smallest scale to enrich the ones derived from the aerial images. Both, the multi-modal features and those of the aerial images at larger scales, are mapped to pixel-wise predictions by means of a decoder. Comprehensive experiments show that the result of the proposed multi-modal classifier are superior compared to those of a uni-modal aerial image classifier; the multi-modal mIOU of 82.3% is 1.4% larger than the one of uni-modal classifier. This demonstrates that aerial image classification can benefit from additional information contained in topographic maps.
AB - Knowledge about past and present land cover is of interest for the assessment of the current status of our environment and, thus, for proper planning of the future. Information on past land cover is exclusively contained in an implicit way in historic remote sensing imagery and historic topographic maps. To make this information explicit, pixel-wise classification methods based on neural networks can be used. The method proposed in this paper aims to automatically predict land cover based on historic aerial imagery and scanned topographic maps. The proposed deep learning-based classifier extracts features at different scales from both modalities and fuses the most complex topographic map features of the smallest scale to enrich the ones derived from the aerial images. Both, the multi-modal features and those of the aerial images at larger scales, are mapped to pixel-wise predictions by means of a decoder. Comprehensive experiments show that the result of the proposed multi-modal classifier are superior compared to those of a uni-modal aerial image classifier; the multi-modal mIOU of 82.3% is 1.4% larger than the one of uni-modal classifier. This demonstrates that aerial image classification can benefit from additional information contained in topographic maps.
KW - Attention-based Fusion
KW - Historical Geodata
KW - Multi-modal Classification
KW - Remote Sensing Imagery
KW - Semantic Segmentation
KW - Topographic Maps
UR - http://www.scopus.com/inward/record.url?scp=105013264735&partnerID=8YFLogxK
U2 - 10.5194/isprs-annals-X-G-2025-221-2025
DO - 10.5194/isprs-annals-X-G-2025-221-2025
M3 - Conference article
AN - SCOPUS:105013264735
VL - 10
SP - 221
EP - 229
JO - ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences
JF - ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences
SN - 2194-9042
IS - G-2025
T2 - 2025 International Society for Photogrammetry and Remote Sensing (ISPRS) Geospatial Week, GSW 2025
Y2 - 6 April 2025 through 11 April 2025
ER -