device. When adjusted, the positions of the stereoscopic 
points indicate the positions of stereo conjugate points in 
the stereoscopic images. Photo 3 is one example. In this 
case, 25 stereo conjugate points are determined manually 
using the 3D display module and are then provided to the 
computer. 
  
Photo 3 Manually determined stereo conjugate points 
In another case, it is also possible to identify stereo 
conjugate lines instead of stereo conjugate points. Photo 
4 is one example of this case. These lines are drawn one 
by one along surfaces such as valleys and ridges, with a 
mouse looking at the stereoscopic model through liquid 
crystal shutter glasses. The steps for drawing these lines 
are as follow. 
1)First draw one line of a stereopair along a valley or a 
ridge on one (left or right) image. 
2)Next draw the second line in exactly the same 
position on the other image. 
3)Repeat these two steps. 
These lines are very useful for automatic stereo matching 
because the accuracy of stereo matching is usually bad 
where valleys and ridges, when an image correlation 
method is used. 
  
Photo 4 Manually drawn stereopair lines 
3.2 Correcting errors of stereo matching 
After the stereo matching, the errors of automatically 
determined stereo conjugate points can be corrected by 
using the man— machine interface easily. Photo 5 is a 
simultaneous display of stereoscopic images and stereo 
conjugate points determined automatically. When the 
stereoscopic model from the stereoscopic images is 
observed, correct points are positioned on the surface of 
the model and error points are above or below the true 
surface. These error points can be corrected easily by the 
mouse, and a precise DEM can be extracted after these 
corrections. 
  
Photo 5 Simultaneous display of stereoscopic images and stereo conjugate points 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B2. Vienna 1996 
  
4.1 
This 
imag 
many 
selec 
Expe 
been 
expl 
One 
such 
route 
appli 
comi 
infor 
on 
deve 
stere 
imag 
get | 
and € 
Phot: 
alteri 
over] 
imag 
and 
oper: 
alteri 
The - 
1) L 
ta 
oL 
(3) 1 
si