oe
4°
che
es
Cor
Che
che
che
vas
of
on-
vas
of
^S.
ing
ity
QUADRANGLE STORAGE
|_| ® QUADRANGLE
IN H.D (6 CHANNELS)
® LOCATING TEST
SITE
TEST SITE ENLARGEMENT [|
VISUALIZATION TO 1:50000 ESCALE ON FOR EACH BAND
MONITOR SCREEN
HISTOGRAM AQUISITION ec
oC | LINEAR CONTRAST
STRETCH
| Selection of max and min
|
L'dues o DN. FF 3
SATISFACTORY
| © Analyses of the gray
| level distribution
je Identification of bands
L for contrast stretch r
ESTABLISHMENT OF MAX AND ß
MIN VALUES OF DN FOR
STRETCHING
p———»
e APPLICATION OF COMBINATION OF
JM DISTANCE BANDS x COLORS
METHOD
Control of intensity !
|
|
| and saturation of |
L_ colors
LEGEND
PLACEMENT
ACQUISITION OF
COLOR PRINT FROM
MONITOR SCREEN
Fig. 5 - Work sequence for the production of color composite images.
between the elements within the group and
by the heterogeneity between elements in
different groups (Curi, 1982). The
specific analysis called ISOMIX (Prelat,
1981) was used because of its great
discrimination power, with the advantage
of being interactive.
The degree of
between the groups is measured by the
similarity of the "centers of groups".
If the distance between two groups is
less than a specified limit, the two
centers of groups are joined and have the
average value of the two original groups.
The process is cyclically repeated until
the standard deviation of each group
center is less than the specified level,
or until the maximum number of permitted
groups is attained. The control
parameters specified by the analyst are:
differentiation
a)standard deviation - controls the
number of classes; b) minimum group
centers and maximum group centers -
controls the category level to be
attained; c) minimum number of pixels per
group - below which it is not possible to
form a new group; and d) separation
threshold - to verify if the classes are
similar or different.
RESULTS AND DISCUSSION
The application of the above
described method of JM distance
indicated, in decreasing order of
efficiency, the following combinations:
254 (2B 5G 4R), 354, and 574,- in the
colors blue, green and red, respectively.
Through visual examinations, we also
considered the following additional
combinations to have informative content:
274, 157, 174, 247, and 154. Tables 1
and 2 show the results of the JM distance
method.
The classification of spectral
information from the cluster analysis
(ISOMIX) permit the union of the soil
units, whether homogeneous or not, as
represented in Figure 4. Table 3
299
presents the control parameters for the
test application.
Table 1 - Rank order of best original TM
bands originais
Bands Results
1.4325
1.2604
1.1288
1.0936
1.0850
1.0682
NWN Od
Table 2 - Rank of the six best subsets of
TM bands
Subsets Results
4 1.4325
45 1.5821
1.6942
2345 1.7624
23457 1.7820
123457 1.8343
N.B.: Rectangles indicate the best
subsets of three or four TM bands.
Table 3 - Control parameters for Isomix
Parameters Values
STANDARD DEVIATION 2,00
MIN GROUPCNTRS 4
MAX GROUPCNTRS 6
MIN PIX/GROUP 50
SEPARATION THRESHOLD 2,50
The delimitations of the soils in
the study area based on the composition
2B 5G 4R are shown in Figure 4.
Comparing this map with the one in Figure
2. reveals that much of the information
coincides, especially for the largest
