Least-Square-Based Joint Coregistration Method for SAR Image Series and Its Application for PS-InSAR Processing

Joint coregistration (JC) approach, based on least square (LS) method, is proposed to improve the synthetic aperture radar (SAR) images' coregistration precision and persistent scatterer interferometric synthetic aperture radar (PS-InSAR) performance. Firstly, the image pairs with small spatial and temporal baselines are used to estimate image shifts to reduce the decorrelation effects. The SAR images are all positioned in spatial-temporal baseline plane and connected with the Delaunay triangulation method. Those arcs with long spatial and/or temporal baselines are deleted with an experienced threshold. Secondly, the shifts are estimated with arc-linked images by conventional coregistration approaches. However, the shifts of arc-unlinked images are still unknown even the arc-linked images are accurately coregistered. For this problem, we estimate the shifts of all the slave images with respect to the reference image from the shifts estimated by the short spatial and temporal baselines. LS method is used to tackle the problem of shifts' error propagation. Finally, all the slave images are resampled according to the estimated LS-method-based images' shifts. 44 ERS-2 images are used to test the performance of JC approach for PS-InSAR processing. The validity is illustrated by the amplitude and phase stability of PS candidates, as well as the deformation performance of PS-InSAR.