Deserts are one of the major landforms on the Earth. While deserts occupy about one-fifth of Earth’s land surface, they have been studied to a much lesser extent. All over the world, desert landforms are expanding ever rapidly and more and more human settlements are finding place in desert regions for habitation. Thus, quantifying and monitoring dunes becomes more relevant from a managerial perspective. Analyzing desert areas using satellite imagery is a challenging task due to weak textural differences and nearly homogeneous spectral responses in most parts of the terrain. In this paper, a post-clustering methodology for change detection of desert sand dunes is proposed. Features based on Radon spectrum are used to cluster dunes of various orientations. These clustered boundaries are used to detect if there are any changes occurring in the dune regions. In the experiments, remote sensing data covering various dune regions of the world are observed for possible changes in dune orientations. In all the cases, it is seen that there are no major changes in desert dune orientations since three decades.
展开▼
