Abstract: Recently, the non-local means filter has been a hot research topic in the image filtering field. The existing preselection based non-local means filters are analyzed intensively, and it is pointed out that they all have defects in terms of feature extraction from image patch. We employ two-dimensional principal component analysis (2DPCA) to extract feature from image patch and propose an efficient non-local means filter. Our contributions to the preselection based non-local means filter are: 1) patch-oriented 2DPCA for extracting features from image patches; 2) automatic selection of the similar sets based on the histogram of similarity distance; 3) local adaptive determination of the similar weight coefficient parameter. Experimental results show that the new method can achieve better filtering results in a variety of images, such as weak gradient image, face image, texture image, and piecewise image.