Research on Building Extraction Based on High-Resolution DSM Images from Oblique Photography

DOI

10.2991/iceep-18.2018.42How to use a DOI?

Keywords

oblique photography, DSM, depth image, gray threshold, area threshold, building extraction

Abstract

Digital Surface Model (DSM) can fully exhibit the undulate features of terrain and landmark, which has obvious advantages in buildings extraction rapidly. The experiments about oblique photography measurements during the M6.5 Ludian Earthquake occurred in Longquan village Longtoushan town on August 3, 2014, in this paper, have been researched. The data are acquired by using the 5-lens oblique photography collection techniques carried by electric Six-rotor Unmanned Aerial Vehicle (UAV), and generating the highly resolution DSM images according to the system model with precision of 4.9cm and elevation of 1494.27-1638.69m. We try to convert the DSM images into the DSM depth images that contain Buildings, ground and trees information. Because in undulate terrain of the mountainous research area, Buildings are constructed according to terrain and all in a muddle. The buildings manifest different floors and coexistence between low buildings and multi-storey buildings, observing from different angles, a serial of buildings form larger architecture covered with dense woods. The threshold of gray values of Buildings and background information displays diversity in the study area, therefore, the method of combining multi-level gray threshold and area threshold is used to extract Buildings for researching in this paper. To obtain the gray values for study area, we transform DSM depth image into binary image by gray histogram method of threshold segmentation, minimum of result is ground, maximum is tree, and Buildings show lower gray values. The gray segmentation threshold of Buildings and background information concentrates on 30 and 45, in this paper, so we select the images of 26, 30, 36, 45 thresholds to binary. The images are combined by threshold of 26 and 36, 26 and 45, 30 and 45, 36 and 45, and we confirm 26 and 45 combination is final result according to actual surface contrast. However, the extracted results also include ground and trees. Buildings, trees and ground show regular otherness in size, so we set up different area thresholds to analysis and extraction for the binary images. At last, we evaluate accuracy by completeness and correctness of extracted Buildings, the result of the assessment is satisfactory. It follows that rapid extracting Buildings by oblique photography DSM images is applicable for not only urban regions but also mountainous areas, which is of great significance in quick acquisition for Building measurements pre- and post-earthquake.

Copyright

© 2018, the Authors. Published by Atlantis Press.

Open Access

This is an open access article distributed under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/).

Download article (PDF)

TY  - CONF
AU  - Xianghua Shuai
AU  - Weiwei Li
AU  - Wei Feng
AU  - Haifang Yang
AU  - Meng Zhen
PY  - 2018/09
DA  - 2018/09
TI  - Research on Building Extraction Based on High-Resolution DSM Images from Oblique Photography
BT  - Proceedings of the 2018 7th International Conference on Energy and Environmental Protection (ICEEP 2018)
PB  - Atlantis Press
SP  - 242
EP  - 250
SN  - 2352-5401
UR  - https://doi.org/10.2991/iceep-18.2018.42
DO  - 10.2991/iceep-18.2018.42
ID  - Shuai2018/09
ER  -