In order to extract significant information for homogeneous extended targets, it is crucial to use speckle filtering in synthetic aperture radar (SAR) images. In this study,a hybrid filtering technique was proposed, which uses both the Sigma filter and the non-local mean filter. The non-local filtering principle is used to select pixels for this method. Pixels with high patch distances that were not similar in the search area were discarded. To estimate the variability of the pixels, the coefficient variation CV is used witch select homogeneous pixels in the search area. According to the results, the proposed hybrid filter improved the filtering criteria in terms of reducing speckle and conserving spatial detail in comparison to Lee filter and the improved Sigma filter. Airborne (Les-Landes, France acquired by NASA JBL) and Space-borne (Sentinel1 C-band hv SAR image of Sousse, Tunisia) synthetic aperture radar SAR data were used for validation.