Quantitavive comparison of phase unwrapping algorithms for SAR interferometry

Interferometric synthetic aperture radar (InSAR) extracts phase information from two complex SAR images to get DEM (Digital elevation map). The technique has been widely used in topography mapping, forest monitoring, and agriculture survey. Phase unwrapping is a key step in InSAR data processing. It directly influences the precision of the final elevation. Up to now, more than 40 phase unwrapping algorithms have been proposed in the world. They can be mainly divided into two categories, one is path- following methods, and the other is minimum-norm methods. In this paper, six classical approaches have been chosen to compare with each other under different conditions based on the criterion of final measured elevation.The work addressed in this paper presents the different effect of six widely used phase unwrapping algorithms under different conditions by comparing the precision of the final measured elevation. InSAR simulation indicates: (1) when the correlation coefficient is great than 0.7, branch cut algorithm has the best effect for the low noise interferogram. (2) when the correlation coefficient is between 0.4 and 0.7, multigrid algorithm can achieve best result for the large noise interferogram. (3) when the correlation coefficient is less than 0.4, the six algorithms all have bad results, and we should use larger filtering window to decrease the difficulty of phase unwrapping at the cost of resolution reduction. The processing of SIR-C/X X-band data of Enta volcano also proves that using different algorithms for different areas can improve the final phase unwrapping result.