Persistent scatterer clustering for structure displacement analysis based on phase correlation network

Abstract

In this paper, we propose a persistent scatterer clustering method for high-resolution structure displacement analysis. Persistent scatterer interferometry can monitor millimetric displacement of structures like bridges, buildings, and roads by analysis at persistent scatterers (PSs), pixels with high coherence in synthetic aperture radar (SAR) images. However, it requires great time and effort to map PSs to the structures in actual geographical space because there are many PSs on SAR images. The proposed method makes this mapping easy by separating the PSs into clusters corresponding to the structures they belong to. The separation is based on a new evaluation index derived from the correlation of phase and distance on a SAR image to consider how likely PSs are on a single rigid structure. The evaluation index is used to construct minimum spanning tree of PSs, and the tree is separated into clusters corresponding to the structures. An experimental result shows that the proposed method can obtain clusters of structures having various shapes and sizes.