Details
| Original language | English |
|---|---|
| Title of host publication | Proceedings of Mid-term Symposium “Innovative Sensing – From Sensors to Methods and Applications” |
| Pages | 347-354 |
| Number of pages | 8 |
| Publication status | Published - 2018 |
| Event | 2018 ISPRS Technical Commission I Midterm Symposium on Innovative Sensing - From Sensors to Methods and Applications - Karlsruhe, Germany Duration: 10 Oct 2018 → 12 Oct 2018 |
Publication series
| Name | International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences - ISPRS Archives |
|---|---|
| Publisher | International Society for Photogrammetry and Remote Sensing |
| Volume | XLII-1 |
| ISSN (Print) | 1682-1750 |
Abstract
Over the past years, the algorithms for dense image matching (DIM) to obtain point clouds from aerial images improved significantly. Consequently, DIM point clouds are now a good alternative to the established Airborne Laser Scanning (ALS) point clouds for remote sensing applications. In order to derive high-level applications such as digital terrain models or city models, each point within a point cloud must be assigned a class label. Usually, ALS and DIM are labelled with different classifiers due to their varying characteristics. In this work, we explore both point cloud types in a fully convolutional encoder-decoder network, which learns to classify ALS as well as DIM point clouds. As input, we project the point clouds onto a 2D image raster plane and calculate the minimal, average and maximal height values for each raster cell. The network then differentiates between the classes ground, non-ground, building and no data. We test our network in six training setups using only one point cloud type, both point clouds as well as several transfer-learning approaches. We quantitatively and qualitatively compare all results and discuss the advantages and disadvantages of all setups. The best network achieves an overall accuracy of 96% in an ALS and 83% in a DIM test set.
Keywords
- Airborne Laser Scanning, CNN, Dense Image Matching, Encoder-decoder Network, Point cloud, Semantic segmentation
ASJC Scopus subject areas
- Computer Science(all)
- Information Systems
- Social Sciences(all)
- Geography, Planning and Development
Cite this
- Standard
- Harvard
- Apa
- Vancouver
- BibTeX
- RIS
Proceedings of Mid-term Symposium “Innovative Sensing – From Sensors to Methods and Applications”. 2018. p. 347-354 (International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences - ISPRS Archives; Vol. XLII-1).
Research output: Chapter in book/report/conference proceeding › Conference contribution › Research › peer review
}
TY - GEN
T1 - Exploring ALS and DIM data for semantic segmentation using CNNs
AU - Politz, Florian
AU - Sester, Monika
PY - 2018
Y1 - 2018
N2 - Over the past years, the algorithms for dense image matching (DIM) to obtain point clouds from aerial images improved significantly. Consequently, DIM point clouds are now a good alternative to the established Airborne Laser Scanning (ALS) point clouds for remote sensing applications. In order to derive high-level applications such as digital terrain models or city models, each point within a point cloud must be assigned a class label. Usually, ALS and DIM are labelled with different classifiers due to their varying characteristics. In this work, we explore both point cloud types in a fully convolutional encoder-decoder network, which learns to classify ALS as well as DIM point clouds. As input, we project the point clouds onto a 2D image raster plane and calculate the minimal, average and maximal height values for each raster cell. The network then differentiates between the classes ground, non-ground, building and no data. We test our network in six training setups using only one point cloud type, both point clouds as well as several transfer-learning approaches. We quantitatively and qualitatively compare all results and discuss the advantages and disadvantages of all setups. The best network achieves an overall accuracy of 96% in an ALS and 83% in a DIM test set.
AB - Over the past years, the algorithms for dense image matching (DIM) to obtain point clouds from aerial images improved significantly. Consequently, DIM point clouds are now a good alternative to the established Airborne Laser Scanning (ALS) point clouds for remote sensing applications. In order to derive high-level applications such as digital terrain models or city models, each point within a point cloud must be assigned a class label. Usually, ALS and DIM are labelled with different classifiers due to their varying characteristics. In this work, we explore both point cloud types in a fully convolutional encoder-decoder network, which learns to classify ALS as well as DIM point clouds. As input, we project the point clouds onto a 2D image raster plane and calculate the minimal, average and maximal height values for each raster cell. The network then differentiates between the classes ground, non-ground, building and no data. We test our network in six training setups using only one point cloud type, both point clouds as well as several transfer-learning approaches. We quantitatively and qualitatively compare all results and discuss the advantages and disadvantages of all setups. The best network achieves an overall accuracy of 96% in an ALS and 83% in a DIM test set.
KW - Airborne Laser Scanning
KW - CNN
KW - Dense Image Matching
KW - Encoder-decoder Network
KW - Point cloud
KW - Semantic segmentation
UR - http://www.scopus.com/inward/record.url?scp=85056153771&partnerID=8YFLogxK
U2 - 10.5194/isprs-archives-XLII-1-347-2018
DO - 10.5194/isprs-archives-XLII-1-347-2018
M3 - Conference contribution
AN - SCOPUS:85056153771
T3 - International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences - ISPRS Archives
SP - 347
EP - 354
BT - Proceedings of Mid-term Symposium “Innovative Sensing – From Sensors to Methods and Applications”
T2 - 2018 ISPRS Technical Commission I Midterm Symposium on Innovative Sensing - From Sensors to Methods and Applications
Y2 - 10 October 2018 through 12 October 2018
ER -