Geodetski list, Vol. 72 (95) No. 2, 2018.
Izvorni znanstveni članak
Earthquake-Induced Building Recognition Using Correlation Change Detection of Texture Features Based on SAR Data
Qiang Li
orcid.org/0000-0002-9559-2170
; Key Laboratory of Crustal Dynamics, Institute of Crustal Dynamics, China Earthquake Administration, Haidian District, Beijing, China
Lixia Gong
; Key Laboratory of Crustal Dynamics, Institute of Crustal Dynamics, China Earthquake Administration, Haidian District, Beijing, China
Jingfa Zhang
; Key Laboratory of Crustal Dynamics, Institute of Crustal Dynamics, China Earthquake Administration, Haidian District, Beijing, China
Sažetak
The detection of building damage due to earthquakes is crucial for disaster management and disaster relief activities. Change detection methodologies using satellite images, such as synthetic aperture radar (SAR) data, have being applied in earthquake damage detection. Information contained within SAR data relating to earthquake damage of buildings can be disturbed easily by other factors. This paper presents a multitemporal change detection approach intended to identify and evaluate information pertaining to earthquake damage by fully exploiting the abundant texture features of SAR imagery. The approach is based on two images, which are constructed through principal components of multiple texture features. An independent principal components analysis technique is used to extract multiple texture feature components. Then, correlation analysis is performed to detect the distribution information of earthquake-damaged buildings. The performance of the technique was evaluated in the town of Jiegu (affected by the 2010 Yushu earthquake) and in the Kathmandu Valley (struck by the 2015 Nepal earthquake) for which the overall accuracy of building detection was 87.8% and 84.6%, respectively. Cross-validation results showed the proposed approach is more sensitive than existing methods to the detection of damaged buildings. Overall, the method is an effective damage detection approach that could support post-earthquake management activities in future events.
Ključne riječi
change detection; synthetic aperture radar; earthquake; correlation analysis; principal component analysis; texture feature
Hrčak ID:
204374
URI
Datum izdavanja:
28.6.2018.
Posjeta: 1.323 *