The priority during speckle reduction in Polarimetric Synthetic Aperture Radar (Pol-SAR) data is to maintain the polarization information. For this reason, it is undesirable to separate each element of polarimetric SAR. Based on the multiplicative noise model and complex matrix, an effective distance similarity measure method is derived, which can process covariance matrix C or coherent matrix T directly. Experiments show that this method can effectively reduce speckle and maintain the polarization information, point targets, and texture structure. Moreover, it can handle large polarimetric SAR with simplicity and ease.