Comparison Of InSAR Phase Unwrapping Algorithm

Phase unwrapping is a key point of Interferometric Synyhetic Aperture Radar(InSAR)technology. The accuracy level of final results in data processing is highly depend on theaccuracy of unwrapped phases. At present, there are many phase unwrapping algorithms. Inall, those algorithms can be grouped into two catagories. The first is the continuous phasesurface by a path-based intergrating operation, and its typical representatives are branch-cutalgorithm, mask-cut algorithm, quality guided algorithm,Flynn algorithm. This algorithmgets unwrapped phases by integrating phase gradients between neighboring pixels in aninterferogram. The second algorithm is using least squares solution, and its typicalrepresentatives are basic iteration, DCT/FFT and PCG algorithm. And this algorithm getsunwrapped phases by minimizing the sum of the square of gradient difference between theunwrapped and unwrapped phase field.This paper introduces principles of several typical phase unwrappine algorithms. Then,this paper does experiments on those algorithms include: Goldstein branch-cut algorithm,quality guided algorithm, least square algorithm and PCG algorithm. At the same time, thepaper compares those algorithms by he follow four aspects:(1)wrapped phase map afterphase unwrapping,(2)comparison to Goldstein branch-cut algorithm,(3)discontinuity map,(4)run time. By the comparison of those algorithms, the paper summarizes the advantagesand dis advantages of those algorithms. Considering the problem that weighted andunweighted least-squares phase unwrapping results are smootuing, an improved algorithmusing phase derivative variance map based on the principle of least-squares phaseunwrapping is proposed. By comparison of the improved algorithm with the weighted andunweighted least-squares algorithm, it is proved that the proposed algorithm can get ahigher accurate and reliable phase unwrapping results.