International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B2. Istanbul 2004 Intern 
6-Estimation of the transformation parameters 
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
The whole procedure of automatic interior orientation (AIO) is 3.5 Pc 
shown in Fig.1. 
Durin; 
extrac 
patch 
matrix 
Resampling the Template negati 
negati 
grey v 
* Figure 3: Desired mask 
Localization 3.6 Es 
3.3 Image pyramid derivation Now | 
1 E the po 
Precise measurement On different pyramid levels we use different representations for sou 
the fiducials. On highest levels we use the whole fiducial mark fau 
including its surrounding(Fig.4 left) On the lower levels the Rrojec 
1 fiducial figure (Fig.4 middle), and only on the lowest level Aca 
Transformation parameters where final measurement is done, the fiducial mark itself is be ki 
used (Fig. 4 right). 
Xe = 
Figure 1: The whole procedure of AIO 
3.1 Extracting image patches J 
To do the job, one doesn't need to use the whole image so we 
just extract the patch including fiducial mark and its 
surrounding (Fig.2). This helps us to reduce the amount of where: 
information and to rise the speed of computation. The size of [a, b. 
patch is dynamically changing in correlation to the image size. 
Ir» 
Figure 4. Left: Pattern of fiducial mark and its surrounding. m 
Middle: Pattern of the fiducial mark figure. Right: Pattern of the 
fiducial mark. 
34D ; : i : 3.7Lo 
3.4 Detection of the orientation of the image 
Hh : RUM ; = ; : From | 
There are eight different possible orientations how to scan an 
image, i.e. wrong reading or right reading with four different SE 
multiple 90° rotations respectively. In general, fiducials are 
normally located on the corners and/or edges of a film as shown 3.71€ 
in Fig. 5. To detect the correct orientation, we analyze the order In this 
of fiducial mark's numbers. pixels 
fiducia 
pyrami 
6 2 5 templa 
the sea 
Figure 2: Extracted image patch neve 
which 
of (r) i: 
  
  
3.2 Resampling the template ] 3 
How to resample the template affects the success and accuracy 
of LSM (Least Square Matching) and output data as more we 
improve the quality of template more accurate output data and 
LSM algorithm will be. Here, since we knew the kind of camera 7 4 8 
so we resampled the template from the extracted patches and 
improved its quality by several pre-processing steps to be used 
by LSM algorithm (Fig.3). 
  
  
  
  
  
  
  
  
  
  
Figure 5. Fiducial distribution on an aerial photo and the 
numbering sequence by USGS of the USA. 
544