A Real-time Selection Method of Adaptive Threshold PS Points for Time Series GB-InSAR Images

Abstract

Ground-based Interferometric Synthetic Aperture Radar (GB-InSAR) is a new type of deformation monitoring tool, which has the characteristics of non-contact, high-precision and all-weather work. With above advantages, GB-InSAR has been widely used in the field of deformation monitoring. The identification of Permanent Scatterer (PS) is the key technology for deformation inversion in the long-term observation process of time series GB-InSAR. Due to the influence of monitoring environment changes, the threshold of traditional PS selection algorithm is difficult to determine, and the adaptability of PS selection cannot be realized that will cause the loss of PS points, thereby reducing the accuracy of deformation inversion. In order to ensure the stability of PS selection when dealing with different environmental conditions, an adaptive threshold PS points real-time selection method for GB-InSAR images based on Gaussian Mixture Model (GMM) clustering algorithm is proposed in this paper. Firstly, the acquired SAR images are grouped, the PS candidate point set is roughly selected based on the amplitude information. Secondly, the PS candidate point set is finely selected according to coherence coefficient, amplitude deviation index and phase stability of each pixel of each group of images, GMM clustering algorithm is used to classify the data three times. This method not only ensures that the number of PS points in each group of SAR images is sufficient, but also improves the stability of the number of PS points. The effectiveness and practicability of the proposed method in this paper are verified by the analysis results of the measured data of the study area.