Level set incorporated with an improved MRF model for unsupervised change detection for satellite images

Abstract

This study proposes the use of a level set incorporated with an improved Markov random field (MRF) model in unsupervised change detection for satellite images. MRF provides a means of modelling the spatial contextual information in the level set, and an edge indicator function is introduced into the MRF model to control the contribution of local information in the boundary areas to change detection. On the basis of the improved MRF model, local label relationships and edge information are considered in the level set energy functional to conduct a novel local term and attract the contours into desired objects. By merging the novel energy term, the proposed approach not only reduces noise but also obtains accurate outlines of the changed regions. Experimental results obtained with Landsat 7 Enhanced Thematic Mapper Plus and SPOT 5 data sets confirm the superiority of the proposed model when compared with state-of-the-art change detection methods.