and
Fg
-hat
sets
1e
an
bec
sifi
art
ppe
d as
sion.
on
istic
the
cene,
to i
eters
nds
ba
ircum
ps
the
s the
1.3 Classification by means of logical multi-valued tree.
The characteristics revealed by means of pseudospectral,
phological and morphometric analysis should be combined so as
to classify every single object as belonging to or not belonging
to the species in question.
Each characteristic (value of area, length of perimeter , pre
sence of mixed ansas, etc.) can be present in various species,
but generally each species is characterized by typical combina
mor
tions of characteristics.
On the other hand certain characteristics by themselves are
not sufficient for the identification of the species, since they
are only indicative: they identify the species when they are co
mbined together.
Establishing and making use of multiple-valued operators, we
establish possible belonging to various classes: i.e. an object
may be an olive tree, a citrus tree or an u.g.Oo. (unidentified
ground object).
Since, in natural images, the objects are not regular, and
not all characteristics, appear completely clearly, it is not
always possible to decide whether an object definitely belongs
to a particular class.
Thus it is necessary not only to define the classes of exis
tance, but also to use different values to give a measure of the
classification obtained in that particular examination, and of
the operators who know to use the various levels.
Thus if we are dealing with olive trees of different ages
(denoted by A and B) along with an unidentified species (U) in
an image, from observation of digitized images we know that the
perimeter of the length between 24 and 5I units can be divided
in three classes: 24-30 in which there are only young olive tre
es and u.g.0.; 31-38 in which A,B and U coexist; and 38-5I in
which A and U coexist.
If, however, we take into account the observation about the
area relative to the same two samples A and B within the 49-205
range, we have:
I) 49-57 B and U
2) 58-II8 A and B and U
3) II9-205 A and U.
adding all this information together we come up with a ta
By
ble called multivalued operator
perimeter
—————
2 3