A support vector domain method for change detection in multitemporal images

Abstract

This paper formulates the problem of distinguishing changed from unchanged pixels in multitemporal remote sensing images as a minimum enclosing ball (MEB) problem with changed pixels as target class. The definition of the sphere-shaped decision boundary with minimal volume that embraces changed pixels is approached in the context of the support vector formalism adopting a support vector domain description (SVDD) one-class classifier. SVDD maps the data into a high dimensional feature space where the spherical support of the high dimensional distribution of changed pixels is computed. Unlike the standard SVDD, the proposed formulation of the SVDD uses both target and outlier samples for defining the MEB, and is included here in an unsupervised scheme for change detection. To this purpose, nearly certain training examples for the classes of both targets (i.e., changed pixels) and outliers (i.e., unchanged pixels) are identified by thresholding the magnitude of the spectral change vectors. Experimental results obtained on two different multitemporal and multispectral remote sensing images demonstrate the effectiveness of the proposed method.