Correlation properties of cascaded recursive median filters

In this paper we will present a series of experiments comparing the correlation characteristics of two nonlinear data smoothing methods. Firstly the Recursive Median sieve, a multiscale dataanalysis system, implemented as a cascade of Recursive Median (RM) filters of increasing window lengths. Secondly a plain RM filter, applied directly to the original input signal. The point that we want to make, backed by a number of experiments, is that the RM filter is not in itself a reliable estimator of location, and it should not be used in data smoothing. As the cascading element in the structure of the sieve, however, the RM filter is very useful. The secondary aim of the paper is to discuss different methods of describing cross correlation characteristics.