A study on image processing techniques for interferometric Synthetic Aperture Radar (SAR) is presented. Phase unwrapping is discussed especially, which is the key problem of interferometric SAR data processing. The interferogram can be calculated by phase difference between two SAR complex images that are acquired by antennas located at different position. Then the phase unwrapping is performed for the interferogram. Finally, according to the geometry relationship of interferometric SAR system, the height of targets can be obtained. The principle, geometry relationship and data processing of interferometric SAR are described, and the system coherence and characters of interferometric phase image are discussed. Noise of the interferogram affected the accuracy of phase unwrapping and Digital Elevation Model (DEM), so reduction of noise is necessary step. The resources of noise and existed algorithms for noise reduction are analyzed, and vector filtering for interferogram is introduced, the efficiency of this method is demonstrated. The area of incoherence in the interferogram embarrassed the automatic phase unwrapping, the data is not reliable in these areas for the low SNR, so we need to extract them as a mask, which will be processed specially in phase unwrapping. A new method of mask extraction from coherence image is presented. The essential difficulty of phase unwrapping is the residues in the interferogram. Different algorithms published recently are summarized and compared, their advantage and disadvantage are analyzed. Phase jump is the obvious feature of interferogram, by which the fringe line image is extracted from filtered interferogram, The interferogram is actually divided into neighboring regions by the fringe lines and the border of the image. A new phase unwrapping algorithm for phase unwrapping based on region has been put forward. The comparison with traditional method shows the advantages of the new algorithm. One post-processing technique for integer variable problem is introduced in order to correct the error of phase unwrapping in some pixels. A lot of application research related to the international cooperative project (ANFAS) has been done. The robust and efficiency of our algorithms are demonstrated by using the ERS-1 and ERS-2 data of Poyang Lake and JingJiang reach pilot sites.
修改评论