Research On High Resolution InSAR Processing Technology

Interferometric synthetic aperture radar(InSAR)has important application value in topographic mapping and military reconnaissance.InSAR technology obtains three-dimensional information of the surface by registering,filtering,unwrapping,baseline estimation,elevation reconstruction,and geocoding,etc.of two or more SAR images acquired from different perspectives.It has great potential in acquiring the Digital Elevation Model(DEM)of the Earth's surface.InSAR image registration accuracy will seriously affect the subsequent InSAR processing and the final DEM product quality.With the increasing development of SAR and InSAR technologies,high-resolution InSAR systems have become the current research hotspots and inevitable trends.Direct application of traditional image registration methods has been difficult to meet sub-pixel level registration precision requirements.In addition,the accuracy of the baseline parameters is also an important factor that directly affects the accuracy of final DEM product,usually in centimeters or even millimeters.For tracking flight mode and repeat track mode InSAR system,it is unable to achieve high-precision inter-satellite baseline measurements,and the baseline accuracy calculated from the radar phase center position recorded by the ephemeris data is far from satisfactory.Therefore this paper focuses on the theoretical and methodological research related to high-resolution InSAR image registration and baseline error estimation.The second chapter analyzes the geometric offset characteristics of InSAR images,which points out that the geometric offset between high-resolution InSAR image pairs has a high correlation with the elevation fluctuation of SAR images.Traditional registration method usually uses a relatively simple mathematical model,and it is difficult to describe the two-dimensional space-variation characteristics of this geometrical offset with high precision.Proposing a registration method that is not affected by terrain is the key to solving the problem of high-resolution InSAR image registration.In the third chapter,proposing a DEM-assisted three-dimensional function method and the geometric method is improved by using the simulated analog image.The geometric method firstly uses the analog image to correct the error between the orbit and the DEM,and then obtains the high-precision registration result through the InSAR geometrical calculation and auxiliary correction of the control point.The three-dimensional function rule directly uses the elevation corresponding to the pixel to accurately fit the registration offset by the approximate linear relationship between the elevation and the registration offset.The experimental results on PALSAR datas show that both methods have achieved good registration results.In the fourth chapter,the InSAR baseline error correction and DEM inversion techniques are studied.Fistly,the accuracy of three baseline estmation methods are compared and analyzed.The baseline estimation method based on the interference fringe frequency,an external DEM assisted baseline estimation method,and a ground control point assisted baseline estimation method.Then the DEM reversion is implemented based on the corrected baseline.The altimetry experiment results of a set of tracking flight mode InSAR data show that the ground control point assisted method has the highest accuracy,the external DEM assisted method is the second,and the interference fringe frequency based method is less effective.The fifth chapter summarizes the paper and discusses the future work.