An order-statistics supported non-linear filter with application to image sequence filtering

A well-known non-linear noise smoothing filter which rapidly adapts to local signal statistics is improved by incorporating robust estimators for the local statistics. The estimators are based on order statistics and adapt, to the local and global properties of the signal. The combination of the filter and the estimators proves to be very tiseful for processing non-stationary signals. The, filter is applied to noise smoothing of dynamic image sequences.