All nine images are read simultaneously into a set of memories. 
For each image the displacement between it and a prior correspon- 
ding image taken within a short time interval is calculated. The 
set of nine displacement vectors form the data for determining 
the change in attitude occuring between the two samplings. The 
time-sequential set of displacement vectors may be used to calcu- 
late the platform attitude variation history, and to generate the 
geometric correction parameters. Data analysis follows the well 
known stereo compilation principles. The effects as seen in 
normal stereo compilation practice are given in Figure 2 (from D. 
H. Alspaugh, 1979). Figure 2 shows the displacement vector compo- 
nents, resulting from translation in three dimensions and three 
Euler rotations (roll, pitch and yaw). 
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
   
   
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
EFFECTIVE TOTAL X COMPONENT Y COMPONENT EFFECTIVE TOTAL X COMPONENT Y COMPONENT 
— — = — — — . e e A Pun N — — om 
-— = cg — — -—- e. . e. by | e I ° ® ° 
— — t— — — — e e . x — A — — — 
l ! l e e 9 ! I Î x | / — e. — 
t | ! o e 9 | l | by | | ! e . « 
| 4 A eie À C ul # 4 à — er 
— e — — . — 9 . e b, — — — «= — — 
ti Et py ps = 
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
Figure 2. The image motion vector set. 
3, Formulation 
3.1 Geometry 
Figure 1 shows the geometry of the problem. Nine square imaging 
devices (32 X 32, 64 X 64, etc. pixel CCD arrays) are placed in 
the focal plane of the instrument. We select the frame of 
reference (x,y,z) tied to the instrument so that the z-axis fis 
perpendicular to the focal plane, pointing to the zenith, x-axis 
pointing in the direction of platform motion and y-axis forming 
the righthand system. The origin of this frame coincides with the 
center of the center array #5. We also select an inertial frame