Abstract: Phase unwrapping is an important step in InSAR elevation measurement，the accuracy of phase unwrapping directly determines the accuracy of InSAR measurement.The unwrapping accuracy of commonly used phase unwrapping methods is susceptible to interferometric noise in low coherence region. In order to solve the problem that the result of phase unwrapping is inaccurate in low coherence region，a phase unwrapping method based on Kriging interpolation and Chan-Vese model is proposed.In this method，Kriging interpolation is performed in the low coherence region， and the interferometric phase values in the low coherence region are recalculated， the interferometric phase values in the high coherence region are unchanged， and a new interferogram is generated.The interferometric fringe boundary of the new interferogram is extracted by Chan-Vese method，and the jump coefficient is calculated according to certain methods， so as to achieve the purpose of phase unwrapping.A mining area in Bange County of Tibet was selected as the test area，the interference phase images which are produced by the pair of sentinel-1A synthetic aperture radar data (C-band) are used as experimental data，the phase unwrapping results of proposed unwrapping method are compared with those of other unwrapping method such as branch cut method proposed by Goldstein.The study results show that proposed method has some applicable conditions，there are little differences between the phase unwrapping results of this method and other methods in the high coherence region.In the condition of low conherence region spanning no more than two interferometric fringes，compared with other methods， the difference between the evaluation inverted from the unwrapping result of proposed method and SRTM1 digital elevation model is within a reasonable range， thus，the proposed method is still suitable.However，in the condition of low coherence region spanning more than two interferometric fringes， the proposed method is not suitable.

Key words: InSAR, Phase unwrapping, Kriging interpolation, Chan-Vese model