Detail-preserving morphological filters in noise suppression

We present an impulsive noise suppression scheme using morphological filters. These filters are composed of directional morphological opening and closing both by reconstruction. A series of structuring elements with different geometrical forms in directional morphological opening and closing are investigated and different masks in grayscale image reconstruction are evaluated. The comparisons in the noise suppression and detail preservation abilities between our scheme and other nonlinear filters are studied. Investigations are also given in the reconstruction abilities by using different structuring elements in geodesic erosion and dilatation.