Research Of Repeat Orbital InSAR Image Registration Based On Multi-scale SPOMF-FMI

For its work mode of all-day, all-weather, space-borne SAR has become one of the most rapidly growing and most effective sensors in the field of microwave remote sensing. As a potential tool for the generation of digital elevation model (DEM) and the measurement of ground surface deformations, InSAR technology make full use of the SAR data. Its principle is extracting ground surface deformation and DEM through the phase difference and space distance between the two complex images from SAR. The processing of InSAR technology consists of image registration, interferogram generation, phase unwrapping and geocoding. As the first step in processing, the registration dramatically influences the accuracy of InSAR products. If there is an obvious offset between the two images for interference, it will cause a serious phase error and the subsequent steps can not be perform. Besides, because the SAR data is complex and its spots noise is serious, the registration for SAR images is far harder than optical images.In this paper, the image registration algorithm in InSAR filed is discussed. First of all, the author makes an analysis for the common registration algorithms in the InSAR technology and sums up the advantages and disadvantages of each algorithm. Then a new automatic registration algorithm is proposed. That algorithm is improved in the base of SPOMF-FMI registration algorithm. In order to prove the advantages of the new algorithm, the author does three experiments with the data from three different sensors. And the results of experiments show that the new algorithm is do superior to the traditional SPOMF-FMI registration algorithm.In addition, the wavelet transform is researched. And the wavelet decomposition is applied in the early period of SAR image registration processing for its characteristics of multi-scale. At the same time, the author proves the effects of wavelet transform in SAR image registration with two experiments:it not only reduces the noise, but also improves the efficiency of image registration.