c) choosing, on the basis of the hystograms, the minimum interv
from
digi als -for each spectral band- able to describe the objects of
e interest
ndivi d) defining and giving the optimum existence volume dimension to
nn cha the objects within the space of the three spectral bands.
ut ca e) spectral exclusion from the image of all the points outside
etric the existence volume
ch it Since the procedure is not sufficient for isolating all and
only all the objects of interest (in our case crowns of olive
trees) an analysis of the form is undertaken, as described below.
y ana (0); (1). 9ee Fig. 2.
mo 1.2 Morphological analysis of remote sensing images.
- The objective is to identify characteristics relative to theform
nd in the objects present in the digitized image, and, in the scene,
zatio some structures (rows, homogeneous zones, etc.).
= These characteristics are identified according to the follow
ing algorithm:
a) with a binary-coded image (one, that is, in which zl is attri
andli buted to the points which come within the significative strip,
Sous and =O to the others) the images are coded according to the
F.M. coding system (2). This system associates every point wi
a r8 th a number, which characterizes the point univocally on the
m basis of its relationships with its visible neighbours. In pa
with rticular the points inside the image are characterized by the
er higest number, and the points defined as 'background' by the
mal zero. In this way the list of the co-ordinates and the F.M.
codes of contour points of the figures present in the images
harac is sufficient for the reconstruction of the whole image.
ral b) a systematic examination of the list of the co-ordinates and
idual of the F.M. codes of the contour points which are gathered
together in significative sets. Each one of these sets may be
characterized by means of an algebric instrument, such as, for
A. example, a grammar and a n L-system. (3);(4).
wai A typical significative set is "CONTORNO", the outline of a
| ang figure, be it full or empty, open Or closed: the outline is
conve the set of visible points of the scene which separates the po
ints of the figure from the other points of the scene which
ysis do not belong to the figure itself. This set is characterized
erge by the property of continuity and congruence. Fig. 3
ey, ve Two meaningful obvious parameters are "AREA" (whole number of
trans pixels of the figure) and "PERIMETRO" (number of pixels of the
ystem. CONTORNO.
olved c) Each of these sets is further divided in subsets of adjacent
points called "syllable"(Fig. 4) the meaning of which is as
grou follows: each syllable is a sequence of vertices belonging to
the same elementary figure (token), belonging to a set which
o the is defined as basic.
A syntetic description of a figure can be obtained from a rec