An Omnidirectional Structuring Elements Adaptive Morphological Filter Based on LMS Criterion

摘要

A new type of adaptive morphological filter is proposed for smoothing an image contaminated with noise in this paper. The filter uses the omnidirectional structuring elements (OSE) and combines linear and morphological open-closing or clos-opening operations. The outputs of the morphological operations by each linear structuring element in OSE are sorted and adaptive weighted processed. The weighted coefficients in the linear combination part of the filter are determined by means of adaptive method under the constrained least mean squared (CLMS) criterion. The new filter is applied to a noisy image and compared with the traditional morphological filters, moving average and median filters. The simulation results have shown that the new filter possesses effective various noise suppression and geometrical structure preservation.