Thesis (PDF) - Signal & Image Processing Lab
Thesis (PDF) - Signal & Image Processing Lab
Thesis (PDF) - Signal & Image Processing Lab
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
Chapter 6<br />
Extrema-Watershed Tree example<br />
Chapter 5 presents a general framework for producing morphological operators that<br />
are compatible to a given tree representations. For every tree representation, a com-<br />
plete inf-semilattice in the tree-representation domain is derived, and a set of mor-<br />
phological operators on that inf-semilattice is obtained.<br />
As for any general framework, the strength and usefulness of the proposed mor-<br />
phological tree-based framework is measured by its ability to:<br />
1) unify existing methods as much as possible, in a simple way, and<br />
2) generate new and useful methods.<br />
The purpose if this chapter is to present an example of a new method that is ob-<br />
tained from the general framework, and to investigate its usefulness. This exemplifies<br />
the strength of the general framework, as a tool for generating new, useful sets of<br />
morphological operators.<br />
Based on the general framework of Chapter 5, all that is needed in order to obtain<br />
a new set of morphological operators is a given tree representation. The more this<br />
tree representation is useful, the more useful these morphological operators are likely<br />
to be, since many of the properties of the tree are inherited by the morphological<br />
operators (like self-duality).<br />
The tree representation selected for the method presented in this chapter is a<br />
particular case of a “Binary Partitioning Tree”, which is a state-of-the-art, general<br />
framework for tree generation, developed by Salembier in [10]. The proposed rep-<br />
resentation is built using an iterative merging process as presented by Salembier,<br />
Garrido and Garcia in [23].<br />
75