Algorithm Research On Image Coregistration Of Sar

In recent years, comparing to traditional means, Interferometric Synthetic Aperture Radar(InSAR) has more salient advantage of speediness, large area observation, full weather, and full time in deriving three-dimensional and deformation information of earth surface, and has became one of the most potential earth observation. Coregistration is a key step of Interferometric data processing. It has important significance in improving the quality of interferogram, realizing the automation of InSAR data processing, and shortening the cycle of data processing by investigating coregistration algorithm against the characters of SAR complex images.Firstly, this thesis introduced the image geometric properties of close compression foreshortening, etc, the storage format of complex data, and indicated the main factors of affecting SAR image coregistration. This thesis emphatically analyzed the conventional coregistration algorithm and process. Based on the characteristics of SAR image, it studied two multi-step coregistration methods, which were integrating correlation and oversampling method and integrating correlation and interpolation of correlation. To test and verify the effectiveness of the two tactics, this thesis chose the SAR image from Naqu, Tibet to perform tests. In the process of coarse coregistration, both the two multi-step methods chose the correlation algorithm based on patches, and provided the initial offset for the follow-up precise coregistration. In the following step of precise coregistration, it respectively chose oversampling and interpolation of correlation to realize the precise coregistration for SAR image, and then compared the precision, reliability, and computing time, etc of the two multi-step coregistration methods.In image processing field, the level set method is applicable to process the change of topology, and it has the advantages of high accuracy, and stable algorithm. This thesis preliminarily discussed the feasibility of applying the level set method to SAR image coregistration. It constructed the function model and data processing, and proved the effectiveness of this method.