| You are in:Home/Publications/Integrating Modern Classifiers for Improved Building Extraction from Aerial Imagery and LiDAR Data | |
Ass. Lect. Haydi Elsayed Elbadawy :: Publications: |
|
| Title: | Integrating Modern Classifiers for Improved Building
Extraction from Aerial Imagery and LiDAR Data |
| Authors: | Haidy Elsayed , Mohamed Zahran, Ayman ElShehaby, Mahmoud Salah |
| Year: | 2019 |
| Keywords: | Building extraction, nDSM, Hybrid system, Image classification, SVMs and ANNs |
| Journal: | American Journal of Geographic Information System |
| Volume: | 8 |
| Issue: | 5 |
| Pages: | 213-220 |
| Publisher: | Not Available |
| Local/International: | International |
| Paper Link: | |
| Full paper | Haydi Elsayed Elbadawy_haidy paper.pdf |
| Supplementary materials | Not Available |
| Abstract: |
This research proposed an approach for automatic extraction of buildings from digital aerial imagery and LiDAR data. The building patches are detected from the original image bands, normalized Digital Surface Model (nDSM) and some ancillary data. Support Vector Machines (SVMs) and artificial neural network (ANNs) classifiers have been applied individualey as member classifiers. In order to improve the obtained results, SVMs and ANNs have been combined in serial, parallel and hybrid forms. The results showed that hybrid system has performed the best with an overall accuracy of about 87.211% followed by parallel combination, serial combination, ANNs and SVMs with 84.709, 82.102, 77.605 and 74.288% respectively. |
