Abstract: As an extension of the Interferometric Synthetic Aperture Radar (InSAR), the multi-frequency InSAR can realize the elevation extraction with high accuracy in the steep terrain. The premise of using multi-frequency InSAR data fusion to realize the elevation extraction in the steep terrain is obtaining high accuracy InSAR interferogram at each independent observation frequency point. In this paper, the local fringe frequencies estimation method based on maximum likelihood estimation, which using in traditional InSAR technology, has been extended to the situation of multi-frequency InSAR, resolving the problem that the estimation result of traditional method which is sensitive to the phase noise, and the result of fringe frequencies estimation can assist to the phase filtering. Simulation experiment results verify the effectiveness of the method in this paper, and the method improves the quality of InSAR interferogram.