| High-precision,fast and large-scale surface deformation monitoring provides important scientific data for geological disaster prevention,geological parameter inversion and law discovery.Multitemporal Interferometric Synthetic Aperture Radar(MT-In SAR)has grown to be an important technology for rapid and large-scale surface deformation monitoring,by the advantages of broad coverage and strong traceability of SAR data.Besides,the technology is not affected by bad weather conditions including clouds,rain,and fog.As the basic input,the quality of interferometric phase affects the density and accuracy of deformation results.Interferometric phase optimization and phase reconstruction are important steps for phase quality improvement.However,in areas with low coherence or vegetation coverage,the existing methods still have the problems of sparse deformation monitoring points and abnormal deformation rate solution.It is difficult to achieve full resolution and high precision deformation monitoring,which seriously limits the potential and applicability of MT-In SAR technology.Therefore,it is necessary to study advanced interferometric phase optimization and phase reconstruction methods,to improve the spatial resolution of monitoring points,increase the deformation details,and enhance the reliability of deformation results.Polarimetric SAR(Pol SAR)data has great potential in obtaining multi-polarimetric information by varying polarimetric modes,to determine the scattering physical characteristics of ground objects.Multitemporal polarimetric In SAR(MT-Pol In SAR)technology is superior to reduce the effects of decorrelation noise,improve the interferometric phase quality with the help of multi-polarization channel fusion or scattering mechanism separation.Among them,the monitoring of distributed scatterer(DS)targets widely existing in natural scenes directly affect the resolution and reliability of the deformation results.Due to the serious decorrelation of DS interferometric phase,MT-Pol In SAR technology improves its interferometric phase quality through three aspects,including spatial constraints,polarimetric constraints and temporal constraints.The corresponding research contents are homogenous pixel identification,polarimetric interferometric phase optimization and phase linking.However,in practical application,the existing methods have not fully explored the polarimetric and interferometric characteristics.In view of the above three aspects,there are still some bottleneck problems to be solved,such as the weak correlation between homogeneous pixels(HPs)and deformation scenes,the poor physical reliability of optimized phase and the unreasonable weighting of observation data during phase linking.Such problems limit the improvement of phase quality and restricts the development of MT-Pol In SAR technology in high-precision surface deformation monitoring.This paper takes the fusion of polarimetric scattering and coherent characteristics into account,solves the main technical problems of improving the interferometric phase quality based on time-series Pol In SAR data.There are three parts for solving such problems,including the multi-scale hierarchical homogeneous pixel identification method based on weighted polarimetric interferometric similarity,the phase optimization method considering scattering mechanism separation and the phase linking method considering time-series scattering consistency.The three parts are progressive and closely related.This paper aims to develop the potential of time-series Pol In SAR data in phase quality improvement,construct a set of MT-Pol In SAR phase optimization and deformation monitoring system,so that it can serve the needs of high-precision and high-resolution fine deformation monitoring.The main contributions and innovations of this paper are as follows:(1)A MT-Pol In SAR homogeneous pixel identification method based on binary partition tree(BPT)is proposed.The co-similarity improves the identification accuracy and enhances the sensitivity of HPs in deformed scenes.The multi-scale hierarchical data structure improves the identification accuracy of HPs in complex scenes.First,considering that the existing methods only takes the polarimetric similarity into account,which is insufficient in deformed area,we propose a new Pol In SAR similarity measure that combines the polarimetric intensity,interferometric coherence and phase.It considers static(polarimetric)and dynamic(interferometric)homogeneity comprehensively,which makes the spatial distribution of the HPs and the deformation phase coincide well.Meanwhile,the co-similarity is simple and the weight is customized,which avoids the adverse effects of large data volume,difficult fusion of multibaseline data,and uncontrollable weights of polarimetric and interferometric information during traditional Pol In SAR similarity calculation.Then,a polarimetric interferometric homogeneous pixel(PIHP)identification method based on BPT segmentation is proposed.Taking advantage of BPT data structure,which can hierarchically express observation information,the SAR image is adaptively segmented into multi-scale homogeneous regions,which is conducive for accurate HP identification in complex deformation scenes.Tested with 30 quadpolarized Radarsat-2 images over Kilauea Volcano,our method shows improvement in phase quality and point density,especially in the deformed areas,and the point density of our method is 4 times higher than that of the traditional method.(2)A phase optimization method for MT-Pol In SAR based on sum of Kronecker products(SKP)decomposition is proposed.It takes the separation of polarimetric scattering mechanisms into account,allows the synchronous decomposition of interferometric phase,improves the coherence of optimized phase and the purity of dominant scattering mechanism,and ensures the physical reliability of results.Considering that the physical interpretation and the result reliability of traditional MTPol In SAR phase optimization method are poor,we use SKP for the expression of observed MT-Pol In SAR data,and use SKP decomposition for the simultaneous generation of polarimetric signature matrix and the structure matrix,which describes the polarimetric scattering process and contains the interferometric coherence information,correspondingly.The performance of our method is evaluated using 30 quad-polarized C-band Radarsat-2 images over Kilauea Volcano.The results show that the proposed method shows good performance in phase quality improvement and point density increasement in five ground object types,including impervious surface,grassland,pasture,shrub,bare land and evergreen forest.The DS point density in the low coherent evergreen forest area is twice that of the existing quad-polarized method.(3)A polarimetric interferometric phase linking method considering time-series scattering consistency is proposed.Using the joint weight with the time-series scattering consistency and interferometric coherence,the proposed method improves the evaluation accuracies of weight matrix for all interferometric pairs,and reduces the estimation errors of phase triangulation process.The weight matrix is the key measure of phase linking method,which determines the participation of each interferometric pair for the singlemaster(SM)phase linking.Existing methods don’t consider the impacts of temporal-changed polarimetric scattering characteristics,leading to large phase closure errors.Based on the polarimetric stationarity,we propose a novel scattering consistency weight measure.Combined with the coherence weight,a joint weight is generated to improve three general phase linking methods.The methods are validated with time-series Radarsat-2 Pol SAR data over Kilauea Volcano,Hawaii.The results show that the proposed methods have smaller RMSE of phase estimation than traditional methods.They can also obtain higher temporal posterior coherence and denser deformation monitoring points than traditional methods.The new methods are essential to the full-resolution and highprecision surface deformation monitoring. |
PDF Full Text Request