pinterregioncontrast

Computes the goodness measure based on inter-region contrast.

Synopsis

pinterregioncontrast [-m mask] [rg_in|-] [im_in|-]

Description

pinterregioncontrast computes the inter-region contrast criterion as defined by M. Levine & A. Nazif^*. This criterion relies on the idea that a good segmentation has a strong contrast between adjacent regions.

This criterion is near 1 when regions are homogeneous and near 0 when regions are heterogeneous.

The inter-region contrast is computed as follows:

    criterion = sum_R [ AiCi / sum Ai ]    Ci = sum [ (lij*|mi-mj|) / (li*(mi+mj)) ]

where

Ai is the area of the region i
Ci is the contrast of the region i computed as follows:
mi is the mean of region i.
mj is the mean of the region j adjacent to i,
li is the perimeter of the region i
lij the length of the boundary between region i and region j.

Caution: Regions with label=0 are not considered for computing.

Inputs

rg_in: a region map.
im_in: an gray level image.

Result

Returns a positive real value in [0..1].
(Use pstatus to get this value).

Examples

Computes the inter-region contrast measure for a simple binarization segmentation process:

   pbinarization 80 1e30 tangram.pan i1.pan
   plabeling 8 i1.pan i2.pan
   paddcst 1 i2.pan i3.pan (label also the background)
   pinterregioncontrast i3.pan tangram.pan
   pstatus

C++ prototype

Errc PInterRegionContrast( const Reg2d &rg_in, const Img2duc &im_in );

Version française

Calcul du critère de qualité basé sur le contraste inter-régions.

Reference

^*M. D. Levine and A. M. Nazif, "Dynamic measurement of computer generated image segmentations", IEEE Trans. PAMI, 7(2): 155-164, 1985.

Author: Régis Clouard