[Insight-developers] new algorithms proposal - semi-implicit level set methods

[Mariana Remesikova]

Hello,
I'm Mariana Remesikova from the Department of Mathematics of the
Slovak University of Technology. Together with my colleagues, we would
like to add several algorithms (classes) to the ITK library. I have
understood that the first step is to propose the algorithms in the
developers list.
Particularly, we would like to implement algorithms for image
denoising, segmentation and counting of the objects in the image. All
these methods are based on level set equations. What would be new, in
our opinion, are the numerical methods for solving the equations. We
have developed a strategy based on semi-implicit time approximation,
that makes the numerical solution unconditionally stable. As a
consequence, there are no severe restrictions for the time step of the
discretization and the methods are robust.
The algorithms have been extensively tested and successfully used
especially in biomedical applications, but they are not limited to
this field. They were described in detail in several papers, the most
relevant is the following

Paul Bourgine, Róbert Čunderlík, Olga Drblíková-Stašová, Karol Mikula,
Mariana Remešíková, Nadine Peyriéras, Barbara Rizzi and Alessandro
Sarti, 4D Embryogenesis image analysis using PDE methods of image
processing, Kybernetika 46 no. 2, 226-259, 2010.
http://www.kybernetika.cz/content/2010/2/226/paper.pdf

For the moment, the algorithms are implemented in C and adjusting them
according to the ITK standard should be straightforward.

Greetings to all,

Mariana Remesikova