Buy article online - an online subscription or single-article purchase is required to access this article.
fast communications
Moving median and center-weighted median filters are employed to remove impulse noise from two-dimensional area-detector data. Two generalized mean filters, which have been used previously in crystallographic data reduction, are compared with median filters. Examination of the smoothed data indicates that the median filters are preferred, in this work, over the generalized mean filters, for data with a high frequency of impulse noise. Like the generalized mean filters, the median filters are useful for eliminating the sharp features (spikes, holes etc.) in the area-detector data while retaining the broader peak-like features.