ul 2004 
<, y) at 
it with 
| point 
o both 
range 
be 1. 
Ww: 
n the 
and its 
ain by 
  
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B-YF. Istanbul 2004 
  
  
  
  
  
  
Z 
4 
i itl 
X,y 
J > 
x (max),y (max) 
Figure 7. Finding the maximum point 
z = ax" +by' +ex+dy+exy+f (3) 
For the area-based template matching, the locating accuracy 
ranges from a half to one pixel, which is, of course, not accurate 
enough for most photographic products. 
4. Analogue Inner orientation 
There was no available photogrammetric instrument in Iran of 
sufficient accuracy that could be used because of the large 
format of the photos (30*30cm). Ways of overcoming the 
problems can be as follows: 1) Making a photographic 
reproduction of the image in suitable pieces, measuring with 
traditional instruments and pinning the pieces together before 
calculation. 2) Shipping the image to a foreign institution which 
has image carriers of sufficient size. 3) Reproducing the image 
; T 2 . 
photographically from the original 30*30cm size to 23*23 
2 : ; 
cm to be measurable in a mono comparator. 4)Using 
overlapping copies (23*30)«" in Planicomp Pl, analytical 
plotter. 5) Scanning the space photos and use the digital 
photogrammetric and image processing systems. Initially the 
third method was used. For determination of geometric 
distortion of camera a grid was used and then the grid and its 
photo were measured. After computation, it was realized that 
geometric distortion due to photography is high (150 micron) 
and the root mean square errors of residuals was 530 micron 
because of large lens distortion of camera, therefore, the first 
method was employed. 
4.1. Pinning the pieces together 
After making a photographic reproduction of the image in 
suitable pieces and measuring with traditional instruments, the 
pieces are then joined together before calculation. The pieces of 
one KFA-1000 photo pinned with conformal using at least four 
common points. The result of pinning the 17957 KFA-1000 
photo being displayed in Table I. 
Points | Vx(micron Vy(micron 
| 1 -] 
1 
zz 
-] 
  
Table 1. Results of pinning 
4.2. Systematic errors of KFA-1000 photos 
The additional parameters of radial distortion of the power of 
five were computed. Film shrinkage is corrected with affine and 
projective transformation. Affine inner orientation was made 
with 4 or 5 fiducial marks, the accuracy of the 17957 KFA- 
1000 photo displayed in Table 2. To take into consideration the 
linear distortion in any direction, a two dimensional projective 
transformation with 5 fiducial marks was employed and the 
accuracy was better than 10 microns. The refraction correction 
is below 2 micron. 
Fiducials Vx (micron Vy(micron 
1 2 
2 
3 
3 
-12 
  
Table 2. Results from the inner orientation 
5. Tests and Results 
The Results of the algorithms implemented can be seen at 
table 3. Input data inculdes KFA-1000 scanned photo, camera 
calibration file including position of fiducial marks in a 
standard conditions of labratory. Output data includes final 
report including positions of fiducial marks in pixel system, 
residual errors of fiducial coordinates, RMSE(Root Mean 
Square Error). Images were scanned on the resolutions of 
micrometer by Intergraph PhotoScan TD. In all tables is 
mean 
the overall accuracy of the automatic interior orientation 
process in micrometer. In table 4. we compare the results 
achieved by using Analogue and Digital method. 
  
Transformation : 21 . 2 Ii 
(D nn excluding (D can including 
fiducial mark 
fiducial mark 
  
  
  
  
  
  
number 5 number 5 
Conformal 25.80 23.65 
Affine 23.84 20.92 
Projective - 10.38 
  
207 
Table 3. The results of using CCF method for localization and 
Surface Fitting for precise measurement