A fuzzy switching filter for removing impulse noise

In recent years, fuzzy set theory has been successfully used in various applications. A typical application area for using fuzzy techniques is to restore images corrupted by impulse noise. In this paper, we present a new framework of removing impulse noise. The most important points is that the types of images are estimated by using the FINDRM and the efficient detail preserving approach (EDPA). When it is estimated that an image has many white and black pixels, the detected noise pixels from the FINDRM are re-checked by using alpha-trimmed means. Oppositely, when it is estimated that an image has a few white and black pixels, the detection results from the FINDRM are used directly. Experimental results show that the proposed algorithm provides significant improvement of PSNR compared with the conventional techniques, and the results are visually very impressive.