Title :
Some statistical properties of median filters
Author :
Ataman, E. ; Aatre, V.K. ; Wong, K.M.
Author_Institution :
Marmara Research Institute, Gebze, Turkey
fDate :
10/1/1981 12:00:00 AM
Abstract :
Some statistical properties of median filters are analyzed. It is shown that median filters can remove impulsive plus Gaussian white noise better than Hanning filters when the amplitude of the impulses is large or the energy of the Gaussian noise is relatively low. It is also shown that, unlike linear filters, median filters preserve sharp changes in signals when the noise energy is not too high.
Keywords :
Convolution; Digital filters; Filtering; Gaussian noise; Nonlinear filters; Signal processing algorithms; Smoothing methods; Speech processing; Tin; White noise;
Journal_Title :
Acoustics, Speech and Signal Processing, IEEE Transactions on
DOI :
10.1109/TASSP.1981.1163659