A Novel Statistically Homogeneous Pixels Updating Algorithm for Long-Term Near-Real-Time Deformation Monitoring

The short-revisit synthetic aperture radar (SAR) provides a large amount of data for interferometric SAR (InSAR) application. Sequential estimator can efficiently utilize the newly acquired SAR images to dynamically monitor surface deformation without reprocessing those already processed images. Sequential estimator uses statistically homogeneous pixels (SHPs) to improve the phase quality. However, the selection of SHPs mainly focuses on SAR images acquired before initial processing, without considering the potential failure of SHPs in the newly acquired SAR images. In fact, the destruction of the homogeneity of SHPs in newly acquired data is a common occurrence, especially in some changed areas, such as the urban construction and the seasonal changes in farmland. The inaccurate identification of the SHPs can compromise the quality of phase filtering, ultimately affecting the precision of deformation results. In this letter, we proposed a novel SHP selection algorithm named SHP update (SHPU) algorithm. SHPU detects the SHPs whose homogeneity gets destroyed in the new acquisitions and updates those SHPs. Experimental results demonstrate that SHPU can detect approximately over 70% of incorrect SHPs and reselect them. The required time of SHPU is only 30% of the time needed for re-estimation using newly acquired data.