- data format manipulations; - radiometric restoration; - geometric 
rectification; - image enhancement; - image compression (decompression); 
- p&ttern recognition; - image annotation. 
Data format manipulations, compression and image annotation. will not 
be discussed here. They are important topics and deserve thorough considera- 
tion in an image processing system, but are of limited relevance in the pre- 
sent context. 
3.1 Radiometric Restoration 
Andrews and Hunt (1977) define image restoration as the techniques to 
remove degrading phenomena from images. Image degradations are numerous 
and familar to photo-interpreters: optical diffraction effects and aber- 
rations, focussing, electronic noise, image motion, atmospheric perturba- 
tions, chromatic aberrations, vignetting etc. Radiometric restoration aims 
at the elimination of these perturbations using a-prior? information derived 
from sensor calibrations prior to actual imaging, or posteriori informations 
derived from imagery already taken of test objects. 
Image restoration serves to relate the image densities to the energy 
used to produce the object's image. It attaches thus a quantitative value to 
the images and is a pre-requsite to quantitative data analysis based on image 
gray values. 
  
  
Figure 1: Aircraft scanner image of Red Mountain, Arizona; (a) with 
coherent noise; (b) after noise removal; (c) after geometric correc- 
tion for panoramic distortion; (d) after low frequency notch filtering 
to reduce shading. 
Filtering may be considered the essential restoration tool. Generally, 
filtering is visualized in terms of spatial frequencies. Noteh filtering, 
i.e. the reduction or removal of specific frequency components, can be 
    
  
    
   
  
   
   
  
  
  
  
  
   
   
  
   
   
  
  
  
  
  
   
   
(Le 
pi: 
SCH 
(Be 
acc 
on 
re: 
ter 
for