oach, 
affine 
CCD 
not be 
uld be 
all. In 
ere the 
eration 
image 
theory 
for the 
h may 
or the 
CCD 
ground 
ntation 
id the 
-to-one 
ces are 
nt free 
biased 
yeneral 
mages, 
by a 
lg six 
of the 
rdinate 
atellite 
inearly 
arizing 
jel and 
lem of 
cussed. 
pective 
xd with 
s, and 
ntation 
letwork 
he nine 
ractical 
many simulation calculations. 
ORIENTATION PROBLEM OF AFFINE LINE-IM- 
AGES 
The basic equations for the three-dimensional analysis 
of affine line-scanner imagery are described in the form 
(Okamoto, et al. (1992)) 
  
  
  
  
Figure-1 : three-dimensional analysis of 
affine line image 
0 = X + DyŸ + Dal, + Da (1) 
Ve = D4Y + DsZ + Dg Xi 
The first equation of Equation 1 denotes the equation 
of a photographing plane in the object space coordinate 
system, and the second equation expresses the 
relationship between the affine line image and an image 
of the object orthogonally transformed into the Y-Z 
plane of the reference coordinate system (X,Y,Z)(See 
Figure 1.). Also, we can see from Equation 1 that the 
three-dimensional analysis of an affine image can be 
separated into the following two processes: the 
determination of the plane including the object and the 
affine image with respect to the reference coordinate 
system and the orientation of the image in the Y-Z 
plane, because the first and second equations in 
Equation 1 have no common coefficients. The 
orientation theory derived by Okamoto et al. in 1992 
can rigorously be applied to the second phase of the 
three-dimensional analysis of overlapped affine images. 
TRANSFORMATION OF CENTRAL-PERSPEC- 
TIVE LINE-IMAGES INTO AFFINE ONES 
characteristics are discussed which are obtained from 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B3. Vienna 1996 
In reality, satellite CCD line-scanner imagery is taken 
central-perspecuvely. Thus, in a rigorous sense, we 
must analyze the imagery based on projective 
transformation. However, such analysis may not be 
effective, because the satellite CCD  line-scanner 
conventionally has an extremely narrow field angle and 
thus very high correlations arise among the orientation 
parameters. This may especially be true when the 
photographed terrain is hilly. In order to overcome this 
difficulty, we will employ the orientation theory based 
on affine transformation by transforming the 
central-perspective images into affine ones. This 
transformation will be explained as follows. 
  
  
Figure-2 : transformation of a central- 
perspective line image into 
an affine one 
Let the ground surface be flat and a central-perspective 
line image be taken at a convergent angle (y (See 
Figure-2.). Further, the image is assumed to intersect 
the terrain at its principal point H.  p(y) denotes a 
real image point and Pg is the intersecting point of 
the ray OAP and the ground surface. The 
corresponding affine image point pa(Ya) can be found 
by drawing the normal to the central-perspective line 
image from Pg. The relationship between the 
central-perspective image point P(Y) and the 
corresponding affine one Da(Ya) is given in the form 
  
zi Ye MM 
7^ bee NN 
Q) 
in Ye VH 
in which c, Ÿc , and YH denote the principal distance 
of the scanner, the measured 1mage coordinate, and the 
principal point coordinate, respectively. The rotation 
angle yy and the interior orientation parameters