:: Volume 4, Issue 4 (3-2017) ::
jgit 2017, 4(4): 33-51 Back to browse issues page
Building Detection in Urban Areas using Features Fusion of Optical and Radar Images in Neural Networks
Maryam Teimouri *, Mehdi Mokhtarzade, Mohammad Javad Valadan Zouj
K. N. Toosi University of Technology
Abstract:   (3199 Views)

In this paper high-resolution SAR, panchromatic and multispectral images are fused for building detection purposes. This fusion is aimed to compensate the defects and shortcomings of these individual data sets. For this reason at first these three data sets are considered individually where some proposed input features are used for building detection. Then these features are fused in different combinations and the results are compared. In all experiments neural networks are applied and their performances are evaluated over different cover types of building roofs. It was discovered that the optimum fusion solution improves the building detection for more then 10% kappa coefficient. Also the proposed fusion strategy caused at least 8 times betters homogeneity of detection results over different roof types. The proposed method enjoying overall accuracy, kappa coefficient and building detection accuracy of% 87.11, % 67.99 and%89.08 respectively confirms the ability to detect buildings of multi-resource radar and optical data.

Keywords: building detection, fusion, radar image, optical image, neural networks
Full-Text [PDF 1752 kb]   (1041 Downloads)    
Type of Study: Research | Subject: RS
Received: 2015/03/17 | Accepted: 2015/10/12 | Published: 2017/04/3

XML   Persian Abstract   Print

Rights and permissions
Creative Commons License This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.
Volume 4, Issue 4 (3-2017) Back to browse issues page