791 
i greatly 
*ed and 
to those 
? system 
iype of 
being 
data are simplified by utilizing only one or two bands at a time, 
and automatic processing is supplemented by user-supervision, 
the problem becomes mathematically trivial. 
A basic characteristic of the ERTS data which can be used 
to classify various objects is the frequency distribution of the 
gray levels in each band. This is conveniently represented as 
a histogram, with the gray levels on one axis, and the number 
of cells in the picture exhibiting each particular response on 
the other, as in Figure 1. The MSS can sense 128 different 
3t particu- 
Lsk, which 
ictive 
5 for re- 
rious 
te" the data 
a.ges using 
le terminal. 
is may be 
t* coordinates 
of these 
gray levels in bands 4-6; however, in any particular scene, not 
all of these may be present. This fine discrimination of gray 
levels, many times that attained by the eye or any photographic 
products, is one of the factors which makes the digital infor 
mation so much more useful for most purposes than the photographs, 
one other being the lack of any degradation inherent in photo 
graphic reproduction. 
Without knowing what objects are in the scene, one can see 
by the histogram which classes of objects are separable by 
response in this band, choose divisions by the valleys on the 
histogram, and use these levels to segmentize the data for 
mapping. In this example (Figure 1), the levels from 1 to 29 
could be mapped with one symbol, those from 29 to 31 with another, 
rjuire re- 
Lrected to 
5 may be 
5 of alpha- 
bations, 
zed equip- 
oftware can 
le experience 
the terminal 
user can 
each case 
nvolved, 
ions attempt 
ative methods 
etc., at the discretion of the user. Of course, it is a great 
help in choosing the intervals to have some knowledge of what 
responses to expect in each band from the various classes one 
is attempting to map. For example, If one wishes to study 
deciduous vegetation, and realizes that deciduous trees give a 
high response in band 7 (i.e., are bright relative to other 
things in the scene), it is easy to reach the conclusion that 
more distinctions should be made toward the upper end of the 
sc&le than the lower end, and that anything below the lowest 
possible vegetational response can be classified as "other". 
This is another great advantage of the digital tapes over the 
photographic product; fineness of detail may be restricted solely 
to the classes in which one is interested, which both emphasizes 
desirable detail and makes for a clearer, simpler picture. 
d use, and 
nalyzing a 
, which 
bject or 
rly sQphis- 
ver, the 
The levels chosen can then be used as inputs to a mapping 
routine, and the process may be repeated for each of the bands, 
the end result being four maps (or more, to emphasize different 
groups in each band), each a gray level "photograph" displaying 
a different aspect of the scene. At this stage, one has an 
analogue of the complete photographic product, which (1) is 
greatly magnified in comparison to the photo, (2) covers only 
the geographic area in which the user is interested, (3) em 
phasizes the details of interest, and (4) is quantitatively 
interpreted.