| Interferometric Synthetic Aperture Radar(InSAR)technology implements the applications of three-dimensional surface reconstruction or micro-deformation detection through the interference information of microwave phase.At present,the number of InSAR satellites in the world is more than ten,but there are still no satellites in orbit capable of serving the stable applications of InSAR technology in China.With the increasement of investments in the field of on-board InSAR in China,researches on InSAR technologies have attracted more and more attention,in which the basic processing technologies of InSAR data are the keys to make full use of satellite data.Recently,China Remote Sensing-29 satellite and GaoFen-3 satellite firstly achieved the interferometric imaging,and it is the pioneer and important experience for China's InSAR satellite development.Interferometric data processing is one of the key technologies from the implementation of interferometric imaging to the ability of interferometry for domestic Synthetic Aperture Radar(SAR)satellite data,and also an important reference from interferometry to the development of normalized,operational InSAR satellites.In this paper,several key technologies of InSAR data processing are studied,especially focusing on exploration and full usage of the interferometric data of domestic SAR satellites.It has been validated that,with the proposed and improved processing technologies in the paper,China Remote Sensing-29 satellite and GaoFen-3 satellite have the initial ability of interferometry and have achieved a relatively high interferometric accuracy.The main contents of this paper can be divided into four aspects:(1)Research on precise co-registration of InSAR images under difficult conditionsSeveral factors,such as the large spatial vertical baseline,the faster spatial parallel baseline,the big elevation of the observation areas,the large difference of repeat frequency of the pulses between two observations,low signal-to-noise ratio or de-coherence,will all lead to the low estimation accuracy of offsets of image pairs,thus causing the difficulties for co-registration of InSAR images.Aiming at the above problems,the qualitative and quantitative influences of geometric and radiation features of imaging on co-registration during InSAR observations are fully analyzed in details.Meanwhile,the characteristics of InSAR image co-registration are discussed and the key points of co-registration are concluded.Based on the above researches,an improved InSAR image precise co-registration method is proposed which takes the relatively large geometric deformation,complicated transformation relation and severe decoherence between images into account.By eliminating the complex no-transform relations between images and then performing automatic locations and extractions of registration points with high quality,an improvement of co-registration accuracy,stability and computational efficiency has been achieved.(2)Research on the improving method of phase noise filteringThe edges of phases and the information of details should be preserved as much as possible when filtering the noises on the stage of InSAR phase filtering.As a time-frequency localized analysis method,wavelet analysis method can effectively distinguish the noises and mutations in the signals.In this paper,the InSAR phase noises are modeled based on the analysis of statistical characteristics and decorrelations of InSAR phases,and an improved multiscale InSAR phase noise filtering method in wavelet is proposed.The InSAR coherence maps are adopted to generate adaptive masks for each wavelet decomposition scale.Experiments proved that the method has good ability of filtering the noises and keeping the phase details and edges.(3)Research on unwrapping method of locally discontinuous and noisy phasesThe noises and jumps of phases are the main factors that affect the accuracy of phase unwrapping.In this paper,the previous researches are firstly summarized.Then,for the difficult unwrapping problems on the conditions of low coherence or discontinuous phases between image pairs,a Markov random field(MRF)model of the interferometric phase is constructed,and then an improved MRF-based and path independent phase unwrapping method is proposed.The method constructs the prior distributions and achieves the energy minimization using Bayesian estimation principle,and thus reconstructing the original phase field,similar to the problem of solving the minimum value of the quadratic function.Meanwhile,the distribution principle of interference phase is adopted in the proposed method.An adaptive energy function is defied for the stochastic model optimization,and then an iteration method of energy function minimization based on gradient descent optimization is designed.The proposed method has overcome the common problems in phase unwrapping,obtained good robustness and stability,and achieved relative good results for the locally discontinuous and noisy entanglement phases that may contain invalid information regions.(4)Experiment and evaluation on altimetry and deformation monitoring of domestic SAR satellites using InSAR technologiesRemote Sensing-29 satellite and GaoFen-3 satellite are the SAR satellites with certain capabilities of interferometric imaging in China's current on-orbit satellites,making them the pioneer and important reference experiences for the development of InSAR satellites.In terms of developing the InSAR satellites,the prerequisite is to form the interferometric imaging,the key is to satisfy the interferometry,and the purpose is to serve the InSAR technologies commonly and operationally.In this paper,the proposed data processing methods are used to conduct evaluation measurement and surface deformation monitoring for the data of domestic Remote Sensing-29 satellite and GaoFen-3 satellite.Experimental results have validated the proposed methods,and verified the capability of domestic SAR satellites in interferometric imaging and interferometry,and evaluated the measurement accuracy at the same time. |
PDF Full Text Request