期刊名称：International Journal of Hybrid Information Technology
印刷版ISSN：1738-9968
出版年度：2012
卷号：5
期号：3
出版社：SERSC
摘要：In former work [18], we propose a scheme to more efficiently preselect similar patchesfor the nonlocal means filter, based on the Patch-oriented two-dimensional principal component analysis (2DPCA) technique. Although the method can yield good results, the computational complexity remains high. For this reason, in this paper we proposed an improvement of the work, which is fast and directly employs features extracted by the patch-oriented 2DPCA to compute the weights. The new approach has been tested on a commonly-used standard test image database. The results demonstrate that our method can significantly improve the denoising effect
关键词：nonlocal means filter; 2DPCA; image processing
