In order to effectively use the temporal information of multi-temporal InSAR and ensure the homogeneity of pixels in the window, a homogenous pixels-weighted interferometric phase filtering method is proposed. The method firstly uses the goodness-of-fit test to identify the statistically homogenous pixels (SHP) in the multi-temporal SAR images, and then the weighted interferometric phase filtering is performed just within those SHP, which ensures that the central pixels is not affected by the surrounding heterogeneous pixels. The experimental results based on real SAR interferograms show that compared with classic Goldstein filter and Lee filter, the proposed filter has advantages in visual effect, phase residues reduction, and phase derivative standard deviation, and achieves the original intention of the design of this filter.