CLUSTER BASED METHOD TO IDENTIFY PERSISTENT SCATTERERS FOR NONLINEAR DISPLACEMENT ANALYSIS OF STRUCTURES

This paper proposes a new persistent scatterer (PS) identification method for temporally nonlinear displacement analysis of structures based on clustering. The identification of PSs, which are pixels of synthetic aperture radar (SAR) images with high signal-noise ratio (SNR), is an important part of displacement analysis technique called persistent scatterer interferometry to ensure the precision. The proposed method identifies the PSs based on that phases of the PSs located on a single structure are strongly correlated. For a reliable identification of PSs, the proposed method separates the SAR pixels into clusters of structures, and evaluates the root mean square of the phase correlations among the members of each cluster. Since this method does not require any models of temporal phase change due to displacement, PSs with nonlinear displacement can be identified. Experimental results show that the proposed method successfully identifies the PSs with nonlinear displacement.