psigmafiltering

Performs sigma filtering on image.

Synopsis

psigmafiltering halfsize eps nbmin [-m mask] [im_in|-] [im_out|-]

Description

psigmafiltering applies the sigma filter algorithm on the input image im_in.

Each pixel is replaced by the mean value of some of its (halfsize*2+1) neighbors. Only neighbors that have a value close to the central pixel are considered for the mean. Close means which difference is lower than the specified parameter value eps. If the number of used neighbors is lower than the specified parameter value nbmin, than the central pixel is replaced par the mean of all its neighbors.

The image border (of size halfsize) is not considered for processing. The output image border is just a copy of the input image border.

Parameters

halfsize specifies the size of the neighborhood. For example, if halfsize=1 then the neighborhood is 3x3 around the central pixel.
eps specifies the maximum difference allowed from the central pixel. It is a grayscale value.
nbmin specified the minimum number of neighbors that can be used to computed the new central pixel value. It is a positive integer.

Inputs

im_in: a grayscale image.

Outputs

im_out: an image of the same type as the input image.

Result

Returns SUCCESS or FAILURE.

Examples

Applies a sigma filter to tangram.pan:

   psigmafiltering 1 20 2 tangram.pan out.pan

C++ prototype

Errc PSigmaFiltering( const Img2duc &im_in, Img2duc &im_out, Short halfsize, Ushort eps, Ushort nbmin );

Version française

Lissage par filtre adaptatif basé sur le choix des voisins.

Author: Régis Clouard