Share Email Print

Proceedings Paper

Village detection based on deep semantic segmentation network in Google Earth satellite images
Author(s): Jiange Liu; Dejun Mu; Jian He
Format Member Price Non-Member Price
PDF $17.00 $21.00

Paper Abstract

Segmenting remote sensing images into no fly zone and open zone is the base of digital geo-fencing construction which is important for the development of unmanned aerial vehicle (UAV) management and control system. The human based segmentation is time consuming, labor intensive and cannot satisfy the real time requirement. Therefore, the automatic segmentation is strongly desired. This study, starting with the rural area in Google Earth satellite images, defines the village as the no fly zone, and other regions are defined as the open zone. The proposed method utilized a deep semantic segmentation network to detect village. A convolution encoder and decoder architecture of SegNet are applied to extract features, and a weighted soft-max classifier, in order to solve the imbalance of sample numbers, is chosen to label the image in pixel level. The results show that it is an effective way to detect the village and exhibit a high accuracy (98%). The method not only can be used as the first step of the automatic construction of digital geo-fencing, but also in the environmental analysis, disaster monitoring of rural area and many other applications.

Paper Details

Date Published: 9 August 2018
PDF: 5 pages
Proc. SPIE 10806, Tenth International Conference on Digital Image Processing (ICDIP 2018), 108065S (9 August 2018); doi: 10.1117/12.2502849
Show Author Affiliations
Jiange Liu, Northwestern Polytechnical Univ. (China)
Dejun Mu, Northwestern Polytechnical Univ. (China)
Jian He, Northwestern Polytechnical Univ. (China)

Published in SPIE Proceedings Vol. 10806:
Tenth International Conference on Digital Image Processing (ICDIP 2018)
Xudong Jiang; Jenq-Neng Hwang, Editor(s)

© SPIE. Terms of Use
Back to Top
Sign in to read the full article
Create a free SPIE account to get access to
premium articles and original research
Forgot your username?