2.1 Orientation 
2.1.1 Interior orientation of digital image 
This program is aimed at determining the transformation from scan- 
ning coordinate system to photo-coordinate system in terms of four 
or more fiducial marks and camera calibration data. In order to com- 
pensate a slight rectangularity and skewing error in digitized images 
a linear affine transformation should be adopted: 
Xg=m() + m4x + moy 
Yg=N) + N4X + N2Y 
in which my ms I, , no» n,» n,denote independent parameters of 
affine transformation. One example of interior orientation is shown 
in table 1. As can been seen from the table, the different scaling 
errors in x- and y-direction are more significant than the skewing 
error, which indicates that the perpendicularity of scanning line 
to stepping direction is very accurate. 
Table 1. Parameters of Interior Orientation 
  
  
  
  
  
Yarameters Left Image Right Image 
m0 m, m, 2141.51 | 1.00506 | -0.00246|2147.42|1.00289| -0.0059T 
ng D4 n, 2157.11.31 0.00257 1.001735|2155.52|0.00410| 1.00162 
  
  
  
  
  
  
  
2.1.2 Relative and Absolute orientation 
vonsidering the results of relative and absolute orientation di- 
rived from two-dimensional digital correlation is less accurate, 
the elements of orientation computed from aerial triangulation have 
been used for SODAaMS directly. 
2.1.5 Computation of parameters for correcting model's deformation 
Although tne distortions in digitized images have been taken into 
acount, the residuals have still effected deformation of stereo- 
moael. From the results of experiments for four mocels listed in 
table 2, it has clear shown that correction for terrain model's 
deformation is essential to increase the accuracy of digital image 
correlation. 
- 675 -