Dept. of Electr. Eng., National Taiwan Univ.SOO-CHANG PEICHIN-LAUNG LEIShih, F.Y.F.Y.Shih2007-04-192018-07-062007-04-192018-07-061998-101350245Xhttp://ntur.lib.ntu.edu.tw//handle/246246/2007041910032491https://www.scopus.com/inward/record.uri?eid=2-s2.0-0032184496&doi=10.1049%2fip-vis%3a19982318&partnerID=40&md5=d1264f8c9306c3c75918727522e3a3c1A new class of recursive order-statistic soft morphological (ROSSM) filters are proposed and their important properties related to morphological filtering are developed. Criteria for specific selection of parameters are provided to achieve excellent performance in noise reduction and edge preservation. It is shown through experimental results that the ROSSM filters, compared to the order-statistic soft morphological filters or other well known nonlinear filters, have better outcomes in signal reconstruction. Two examples are given for demonstrating the flexibility of the proposed filters in signal processing applications. © IEE, 1998.application/pdf1225926 bytesapplication/pdfen-USImage reconstruction; Morphological filtering; Recursive order-statistic soft morphological filtersEdge detection; Noise abatement; Parameter estimation; Performance; Recursive functions; Signal filtering and prediction; Statistical methods; Morphological filtering; Image reconstructionjournal article2-s2.0-0032184496http://ntur.lib.ntu.edu.tw/bitstream/246246/2007041910032491/1/00741946.pdf