An efficient compressive sensing based PS-DInSAR method for surface deformation estimation

Abstract

Permanent scatterers differential interferometric synthetic aperture radar (PS-DInSAR) is a technique for detecting surface micro-deformation, with an accuracy at the centimeter to millimeter level. However, its performance is limited by the number of SAR images available (normally more than 20 are needed). Compressive sensing (CS) has been proven to be an effective signal recovery method with only a very limited number of measurements. Applying CS to PS-DInSAR, a novel CS-PS-DInSAR method is proposed to estimate the deformation with fewer SAR images. By analyzing the PS-DInSAR process in detail, first the sparsity representation of deformation velocity difference is obtained; then, the mathematical model of CS-PS-DInSAR is derived and the restricted isometry property (RIP) of the measurement matrix is discussed to validate the proposed CS-PS-DInSAR in theory. The implementation of CS-PS-DInSAR is achieved by employing basis pursuit algorithms to estimate the deformation velocity. With the proposed method, DInSAR deformation estimation can be achieved by a much smaller number of SAR images, as demonstrated by simulation results.