Abstract: Non-local means have been applied to the field of polarimetric SAR image filtering. However the traditional methods process SAR data with multi-look before filtering, which reduces the resolution of SAR images with enormous computation costs that severely restrict non-local means filteri9ng performance. So a fast non-local mean filtering algorithms is proposed for single-look polarimetric SAR images. A new non-local means model was reconstructed in accordance with the multi-view concept to seek power, and single vision remodeling, then combined with KL divergence. Fast computation in the new algorithm was implemented with an integral image. Finally, the proposed algorithms were validated with a Radarsat-2 fully polarimetric SAR image. Experimental results show that the new algorithm not only improves the filtering effect but also significantly enhances operational efficiency.