A signal dependent rank order median weighting filter combined with fuzzy theorem for impulse noise suppression

• Main Authors: Chun-yi Lin, 林峻毅
• Other Authors: Shiau-fang Lei
• Format: Others
• Language: en_US
• Published: 2007
• Online Access: http://ndltd.ncl.edu.tw/handle/64202392020001962391

碩士 === 國立成功大學 === 電機工程學系碩博士班 === 96 === In transmission an image in channel, due to a limited of bandwidth, if the channel is so poor that the noise variance is high enough to make the threshold operation done at the front end of the receiver will contribute to impulse noise on images. Traditionally, median filter are used to suppressing this kind of noise but induced the damage on image details. An ordered statistics filter based on switching scheme is popularly used to improve the damage on image detail when we are de-noising; the central concept of the ordered statistics filter based on switching scheme is by use the characteristic of surrounding pixels to find out the most similar value for replace the noisy pixel. The SD-ROM[3], SM[18], New Efficient[31], Fuzzy filter[32] and MMEM[10]…etc. The proposed method is based on the structure of New Efficient[31], also including the fuzzy concept in order to improve the difficult to get threshold when the difference of noisy pixel and surrounding pixels is not large, and an error judgment may occurred because of the misfit threshold when processing image details and thin lines. Theses situations may cause the poor performance on reducing images. By judgment the existence of impulsive noise is expressed by fuzzy rules for improvement, and referenced the MMEM filter[10] for improve the salt and pepper noise. By combine these methods to get better performance than the New Efficient algorithm. Simulation results can show the improving part in images and also on PSNR and COR value.