WEIGHTED ADAPTIVE LMS L-FILTER BASED ON CORRELATION COEFFICIENTS (Lc-WOS)

Róbert Hudec - Stanislav Marchevský

   The weighted adaptive order-statistic LMS L-filter (L-WOS) based on the analysis of correlation coefficients (Lc-WOS) is presented in this paper. This filter is a combination of an L-filter and a weighted order-statistic (WOS) filter. The multiplicity of each pixel from an input observation is obtained by correlation relationship between sorted pixels grouped around this pixel and sorted pixels grouped around the central pixel in the input observation. This method can serve to identify the noise model in unknown images and at the same time an optimal weight mask and weight coefficients can be used to its elimination. Gaussian white noise and the mixed (Gaussian plus impulsive) noise studied the adaptation properties of the Lc-WOS filter.

Keywords: adaptive LMS L-filter, weighted order-statistic, correlation coefficient

[full-paper]