,
I. MEUNIER 
,
and J.C. KLEIN
,
I. MEUNIER ,
and J.C. KLEIN
:
Centre de Morphologie Mathématique, Ecole des Mines de Paris
35, rue St-Honoré 77305 Fontainebleau, France
zana@cmm.ensmp.fr, klein@cmm.ensmp.fr, fax: (+33) 01.64.69.47.07
and
: Eye University
Clinic of Créteil, Paris XII
40 avenue de Verdun 94010 Creteil, France
fax: (+33) 01.45.17.52.27
Abstract.
We present a region merging algorithm based on
a watershed cell decomposition and on morphological treatments for macula
recognition. After noise removal, a watershed is performed to produce a
mosaic image. The resulting decomposition segments the image in cells while
maintaining the shape of the Macula. A cell inside the macula is selected
as the first step of a merging algorithm. Morphological filters are performed
to enhance the capillaries surrounding the macula. A complex criterion
based on the gray values and of edges of the filtered image is calculated
to merge the cells of the macula while rejecting perifoveal intercapillary
zones in order to produce the contour of the macula.
keywords:
Region merging, Mathematical Morphology, Watershed, Image
understanding, Ophthalmology.
ask for a Postscript version of the article (508Ko)
| More about authors: | ||