Research On High-coherence Point Selection Method For Multi-application Ground-based Differential Interferometric Radar

Carrying out monitoring and early warning for monitoring areas with landslide hazards is an important part of disaster prevention and mitigation.Before the macroscopic instability of the slope body occurs,the surface of the slope body usually undergoes slight deformation.GB-DIn Rad(Ground-Based Differential Interferometric Radar)is a high-precision deformation measurement device,which has been widely used in the field of deformation monitoring.High coherence point selection is an important process in the realization of deformation measurement based on phase-difference interferometry.This paper mainly focuses on the high coherence points selection methods for different application scenes to solve the shortcomings of conventional high coherence points selection methods.First,for rock slopes,there are too many high-coherence points selected by conventional methods,which is inefficient,in addition,the selection of high coherence points based on a single method is easy to achieve,but it is also prone to misselection,resulting in the accuracy of deformation measurement in the later period.Secondly,for vegetated mountain slopes,the quantity of high coherence points selected by conventional methods cannot meet the requirements of high precision deformation measurement,which is not conducive to the later deformation analysis.Finally,using GB-DIn Rad for real-time deformation measurement requires the selection of high coherence points based on a grouping scheme,and the phase quality and quantity of high coherence points in each group have serious dynamic changes.Based on GB-DIn Rad,this paper proposes a multi-application-oriented 11 selection method.The specific work is as follows:(1)A multi-stage selection method of high coherence points for rock slopes is proposedIn order to effectively solve the problem of insufficient precision and inefficiency due to the excessive number of high coherence points selected by single amplitude dispersion method in the rocky slope scenes,a multilevel selection method of high coherence points based on amplitude dispersion and time series coherence coefficient is proposed.The basic idea of this method is: firstly use the amplitude dispersion method to classify the pixels,secondly use the cross-correlation coefficient and the time series coherence coefficient to perform the second-level and third-level selection in turn,finally perform the fourth-level selection based on the phase information,and filter out the final High-quality pixels,that is,HCP(High Coherence Points).The experimental results of an open pit show that the proposed method can effectively solve the problems of precision and efficiency of conventional methods in selecting high coherence points in high coherence scenes such as rock slopes.(2)An improved Sta MPS method for vegetated mountain slopes is proposedIn order to effectively solve the problems of insufficient quantity of high coherence points selected by conventional methods in vegetated mountain slope scenes,an improved Sta MPS(Stanford Method for Persistent Scatters)method is proposed.The basic idea of this method is: firstly,the set of candidate high coherence points is obtained by using the amplitude dispersion method.Secondly,spatial dimension filtering is carried out based on the phase of the pixels,and the coherence coefficient is estimated for the filtered interference phase.Finally,non-high coherent points are used to estimate the coherence coefficient threshold,and DS(Distributed Scatterer)selection technique is introduced to further increase the number of high coherent points.The experimental results of a typical vegetated mountain slope show that the proposed method can effectively solve the quantity problems of high coherence points selected by conventional methods in low coherence scenes such as vegetated mountain slopes.(3)A high coherence point selection method for real-time deformation processing is proposedIn the real-time processing of time-series radar images,a grouping scheme is generally used to select high-coherence points.At this time,if the amplitude dispersion threshold is fixed,the phase quality and quantity of high coherence points in each group will undergo serious dynamic changes.Aiming at the above problems,this paper proposes an improved adaptive adjustment method of the amplitude dispersion threshold.The basic idea of the method is as follows: firstly,the radar images are grouped,then the abnormal images are removed,finally,the amplitude deviation threshold is adaptively adjusted by using the fluctuation factor of amplitude dispersion and the number of high coherence points as the conditions of adjusting the threshold.The experimental results on the measured data of two different monitoring scenes show that the proposed method can select stable high coherence points in different scenes and different time periods,effectively ensuring the real-time processing accuracy of ground-based differential interferometry technology.