International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part Bl. Istanbul 2004 
  
r 
A, 0 0 By Gi 1 | Xg4 70i) : 
Aen 0 0 Ben Gen Xigox Yea Yen) 
0 A 0 Bg Gg Xigex Xig4 7g.) 
Feliu oue oda. 
0 Aga 0 Ben Gen in A En mU 
0 0 Ag Bg C X bsa | Med m 
L 0 0 Aiga Big, Giga ons | £a 7e; )| vd 
  
  
  
  
: : T . 
According to the least square principle: V^ x/ = min, so 
j aforc 7 
the unknown parameters X e, , X yg i, X que uu, X, and 
X,, can be solved by the following iterative process: 
Xeon i zl 
À.| |4 00 8 Gj400 8 G[4 00 8 G| x, | 
Aw.FjO04 0 8 GO A4 08 G/((04 08 G| À 
Xe 004 B,GJl0 04 B G0 04 B GDigainn)] 
   
3.3 Calibration contents in multi-camera system 
Though the principle and technique of CCD cameras is quite 
different from that of common analog cameras, the two 
categories are similar in calibration specifications and method. 
However, the calibration contents vary from different purposes. 
As for our multi-camera system, the following items are 
calibrated first : The inner elements of each camera: principal 
point( x, , yg )and principal distance(f);The optical distortion 
parameters : radial distortion factors( Æ, , &, ,...) and eccentric 
distortion factors( p, , p, ). 
After the inner elements and distortion parameters of each 
small-sized CCD camera has been obtained using the solution 
of space resection with multiple images, the cameras are 
mounted onto the platform specially designed for 
multi-camera combination system and they are electronically 
controlled to take the images of the same calibration field 
simultaneously. With these data, real exterior elements: linear 
elements( .X , , Y; , Z, ) and angle elements ( 9 , & , K ) of the 
multi-camera system, relative exterior elements of each 
camera to the principal optical axis of equivalent image 
(0.,0,,0., d$, dS, dS. ) are to be determined. 
3.4 Multi-camera system correction 
The following corrections are indispensable for practical system. 
The corrections are based on system calibration results. 
1) Exterior elements correction 
The correction component of exterior elements 
(dS, d$, dS. , dO y ,dOy, dO; ) is the difference between 
system design value and its corresponding calibrated value. 
2) Image coordinates correction 
Denote corrected image coordinates as (x,y), observed image 
coordinates as ( x;,,, V;, ).radial distortion factor as k, . (Since 
our multi-camera system use small-sized non-metric CCD 
cameras, we can take the radial distortion factor Æ, into 
consideration only.) Suppose 7r is the radius from image point 
to principal point. Then correction model for image correction 
is: 
PEE C ef = = 
E =2m = Xo Tl 7 Xo)! 
Yt , — , — , À 
X -m—y m Yo t Ki Cy im Yo )r 
Where, r? zx Ex + (Vim — y)” 
un 
3) Equivalent image correction 
The equivalent image coordinates (x , y) is calculated on the 
basis of (x'y'). Further, (x , y) needs to be corrected by 
correction component of exterior elements. The correction 
equation is: 
; 
X N Xy 
dx z ds y ys +f + pi * Wx * yd; 
ex diet Ads; rf do, + 77.46, t xd6; 
4. EXPERIMENTS 
The following experiments are carried out to test and verify 
the above mentions. 
Four FUJIFILM FinePix 4700zoom areal CCD cameras are 
used in the experiments. FinePix type is characterized by small 
size, handy use, high light sensitivity, large storge 
etc..especially, the capabillity of locking principal distance. 
3 and 4 FinePix 4700zoom cameras are mounted on 
experimental multi-camera platform to construct the 3-camera 
combiantion system (Fig. 12) and 4-camera combination 
system (Fig. 13). 
   
Fig. 12: 3-camera system Fig. 13: 4-camera system 
Table 1 gives the result for each single camera 
calibration. The cailbration is done through outdoor 
  
  
     
   
    
     
    
    
   
    
   
   
    
     
   
    
    
    
   
   
     
   
   
    
    
   
   
    
  
    
      
    
Interna 
calib 
resec 
  
Notes 
1. The 
degre 
2.The 
conve 
3.The 
Thec 
also d 
electr 
time. 
in spa 
Thef 
Acad 
3-can 
Fig.1: 
equiv 
mosai