Heterogeneous change detection with PRISMA and COSMO-SkyMed Second Generation imagery for natural disaster management

Abstract:
Change detection (CD) is among the most important tools in natural disaster monitoring. Special emphasis is on heterogeneous CD methods, which allow for a faster response. In this paper, we propose a novel heterogeneous CD method tailored at working with image domains of very different dimensionality, which allows for a greater applicational flexibility. The proposed method integrates deep image-to-image translation, spectral clustering concepts, and manifold learning, and works in a fully unsupervised manner, further enforcing a fast implementation in real-world scenarios. From an application-oriented perspective, the focus is on the recent PRISMA and COSMO-SkyMed missions of the Italian Space Agency.