d)
- 494 -
tangle by subtracting according to a defined multi-valued dif
ference operator from a rectangle all the elementary figures
associated with the syllables. (5). Fig. 5.
The elementary figures selected are those shown in Fig. 3 and
are divided in two sets of figures which are geometrically e
qual but distinguishable one from the other for the fact that
those in the first set are full, while those in the second a
re empty.
the syllables or sequences of adjacent syllables identify sets
of significative points with respect to a description;e.g.
ANSA i (BIGHT i): sequence of vertices which belongs to the
same empty elementary figure.
PUNTA i (PEAK i): sequence of vertices which belong to the sa
me full elementary figure.
SCALINATO (STEP-WISE): sequence of n vertices which belong al
ternately to a full figure and to an empty figure and which
allow the identification of the complexity of these sets in
function of how may primitive elementary figures are required
for their formation.
The sets of points thus constructed define classes of convexi
ty or concavity within the single object, these assuming diffe
rent meanings according to the context of both the experiment
itself and the instrument used for it.
Thus, for example, STEP-WISE (SCALINATO) is a set which can
represent a straight line in which vertices are included bec
ause of the digitization. If tis happens, it will be classifi
ed as a straight line; if it does not, that is, if some part
of the experiment or the surveying instrument stops it happe
ning, STEP-WISE will denote a non-rectilinear line.
In our experiment the SCALINATO shape has to be considered as
characteristic of the crown and not due to the A/D conversion.
In Fig. 3 a case is shown; in Fig. 6 an italian description
of the same object is given where the mentioned characteristic
of the object are identified; in Fig. 7 reconstruction of the
object and two kind of baricenters are described.
A similar morphological analysis can be extended to the scene,
treated as one single object. This means the possibility to i
dentify each single object with well-defined set of parameters
and finding out the relationship within these sets.
Typically, it was decided in our work to explore three kinds
of sets: (a) the baricenter of the binary figure; (b) the ba
ricenter of the original figure; (c) the rectangle that circum
scribes the single figure.
The aim is that of identifying the reciprocal relationships
the objects of the images which allow us to discriminate the
families of objects: for example, in the case of the trees the
distance between each one seems to be a powerfull one.