The Registration Algorithms Of Space-borne Interferometric Synthetic Aperture Radar Images

The registration of the complex space-borne SAR images is a key point of Interferometric Synthetic Aperture Radar(InSAR) technology,and is also the base for the radar interometry technique to realize specific applications,while the select of high precision and high reliable control points is an important segment to realize precise registration of InSAR images.At present,the average fluctuation function of phase difference image method, the maximum interference spectrum method and correlation coefficient method are mainly used in the registration of the complex space-borne SAR images.These three kinds of methods are to select the control points based on a registration measure function, which are of high degree of automation, and are insensitive to the rotation and brightness changes to a certain extent.So they are applicable for some areas, but in contrast, they can’t meet the requirements of SAR image registration in the weak edge or single andobvious texture featureareas whose coherence is low, because there are often some gross errors and much more intensive registration points at one area in practical application.First,this paper introduces some common registration algorithms, thenuse intensity cross-correlation algorithm and coherence algorithmrespectively to complete the registration experimentsbase on the GAMMA software platform.This paper compares the two algorithms mainly from the sides of the number of residual points, the registration error and computational efficiency, summed up the advantages and disadvantages of each method in the end.Aiming at the problem of gross error may exist in the control point, this paperapplies the overall variance factor test for gross error detection that the control point file contains; this paper uses studentized residual and the prediction residual to reject the gross errors respectivelythrough the establishment of trend surface model parameters. In the end,the LS is used in the paper to evaluate theaccuracy of the remaining control points until it meets the SAR images registrationrequirements.The experiments result verifies the feasibility and robustness of this method.