Abstract: A new speckle filtering algorithm based on four-component scattering model was proposed in this paper.Firstly,the four-component decomposition was applied to get scattering category and span value of each pixel.Secondly,polarimetric and spatial similarity measurement was utilized to select honogeneous area.Finally,the LMS filteing was performed based on local statistical properties of homogeneous area.Experimental results using L-band fully polarimetric SAR data acquired by AIRSAR sensor show that the proposed algorithm not only filters speckle effectively but also preserves polarimetric and spatial information well.