36
IrIs BIometrIC modeL
2.7.3 Classification Method
e classification consists in grouping similar elements of E (E ≠ ∅).
Each group represents a nonempty and homogeneous subset. All
these groups cover the set E.
We can mention two main classification methods (Saporta 1990):
1. e hierarchical classification leading to merge in the same
group two elements of E that have a given level of accuracy,
while at a higher level of accuracy, they will be distinguished
and belong to two different subgroups
2. e nonhierarchical or partitioning classification, leading to
decomposition of the set E of all elements into m disjoint sets
or equivalence classes whose number of classes m is fixed
A hierarchy Η is called indexed if there is a function ind: Η → ℝ
+