This paper mainly focuses on the 2-D phase unwrapping algorithm for Interferometric Synthetic Aperture Rader （INSAR）. The statistical property of error in the phase differences of INSAR （Interferogram） is firstly presented in brief

which shows this error could be assumed to be Gaussian noise. The algorithm of least-square 2-D phase unwrapping using FFT（LS-FFF） is then discussed

and it is proved to have a good property of robustness in Gaussian noise. Finally

the idea of parallel processing for LS-FFT is put forward

and is verified with the computer simulations.