cultural items than any other single frame (photographic) of the 
same area. This is true at least with respect to paddy and sugar- 
cane. In effect, the color infrared frame contains multiband in- 
formation. 
Consequently, with respect to the processing of color infra- 
red frames, we expect the automatic interpretation system to 
(i) have the capability to distinguish between colors and 
between textures ; 
(ii) make use of contextual information to identify and cla- 
ssify objects otherwise difficult or impossible ; 
(iii) see the image as whole, not sequentially, :pixels by pi- 
xels ; and, equally important, 
(iv) be cost effective, especially in countries where traine 
ed man-power is cheap. 
However, it seems that such a system is a thing of the future.As 
it is, perhaps the only superiority of existing automatic inter- 
pretation systems over the human visual system is that multiple 
band images (available separately) of the same area can be simul- 
taneously processed. It should also be added here that no results 
are yet available on the best choice of filters for panchromatic 
films so as to derive information on agricultural crops, equal 
to / more than that possible through the use of the color infra- 
red film. 
What follows is a feasibility study, not a definitive des- 
cription, of the digital analysis of color infrared frames. 
3. Digital Processing of Color Infrared Frames 
(a) Measurement of features : The color scanner (slides No.l 
and No.2), fabricated at the Indian Institute of Science, is used 
for this purpose. Some relevant technical details are as follows: 
Picture Resolution : 100micrometers 
Maximum picture (print) 
size for scanning 3 25.4 x 30.5 cm 
Rate of scanning : A print of size 
129.9.x 12.9 cm. can 
be scanned in 15min, 
Number of scan lines : 256 lines per in: 
The color infrared frame under study is converted to an en- 
larged color print (by standard procedures), and the print is 
mounted on the drum of the color scanner. Naturally, there is 
some loss of information (and,possibly, an introduction of dis- 
tortion) in this transformation of a transparency to a color 
print. However, this is not taken into account in this feasibi- 
lity study. 
See Fig.l for 2 block schematic of the color scanner(Slide 
No.2). A small portion of the picture to be scanned is illumi- 
nated by & light source via fibre optics light guides and focus- 
sed onto the image plane of the optical system. The reflected 
beam passes through & prism to give the spectral components which 
are then passed through the (primary) color filters, Wratten Nos. 
47, 58, and 25 (blue, green, and green respectively). The three 
filter outputs are sensed by three different photosensitive devi- 
  
   
  
  
  
  
   
  
  
  
   
  
  
  
  
  
   
   
   
   
  
   
  
  
   
  
    
  
  
   
  
  
  
    
  
   
  
  
  
   
  
   
    
  
   
  
  
   
    
£g HH 0 
f0 ct H xf He ct