Unsupervised Change Detection in Multispectral Remote Sensing Images Via Spectral-Spatial Band Expansion

Sicong Liu,Qian Du,Xiaohua Tong,Alim Samat,Lorenzo Bruzzone
DOI: https://doi.org/10.1109/jstars.2019.2929514
IF: 4.715
2019-01-01
IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
Abstract:Due to the limited number of spectral channels in multispectral remote sensing images, change information, especially the multiclass changes, may be insufficiently represented, resulting in inaccurate detection of changes. In this paper, we propose to use unsupervised band expansion techniques to generate artificial spectral and spatial bands to enhance the change representation and discrimination for change detection (CD) from multispectral images. In particular, in the proposed approach, two simple nonlinear functions, i.e., multiplication and division, are applied for spectral expansion. Multiscale morphological reconstruction is used to extend the band spatial information. The expanded band sets are then used and validated in three popular unsupervised CD techniques for solving a multiclass CD problem. Experimental results obtained on three real bitemporal multispectral remote sensing datasets confirm the effectiveness of the proposed approach.
What problem does this paper attempt to address?