projection 
ne control 
 ¥, ) . The 
y factor |. 
ous optical 
(6) 
ue, while 
ip between 
ors include 
ective-lens 
' with the 
oint that 1s 
rors caused 
ie test done 
| according 
the image 
is shown as 
ce with the 
ortion. The 
e frame (it 
ollimation). 
tion of the 
) the center, 
ot zero. 
be created 
; created by 
the digital 
J" ) is pixel. 
s Figure 5, 
re 6. 
Istanbul 2004 
  
  
136 
150 
Fig. 
Fig.6. Distortion primary model for Y 
By using ordinary camera calibration methods, the 
  
  
  
  
chart on the CCD 
We n 
| 
= 
  
  
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part BS. Istanbul 2004 
observation value is limited for the distortion factors (such 
.as k, 3 k, » P1 » D; ) The relationship between the 
unknown factors will have an effect on the measure quality 
of the distortion coefficients. However, with the method 
discussed in this paper, the measure result directly reflects 
the real status of distortion, because only imaging system 
error values are to be measured with several hundreds of 
observation values. For real applications, according to the 
known distortion error values on the “dispersed points”, 
establish distortion superior model by interpolating and 
computing the distortion of all pixel positions; or by 
temporarily interpolating the distortion of a certain image 
point. The interpolation principle of the two methods is 
consistent. The required storage capacity should at least be 
equal to the resolution of the chip, i.e., several G or tens of G, 
in correspondence with the distortion superior models on 
each focus plane. The method of temporary interpolation is 
of more benefit. 
With regards to the interpolation of the distortion error of a 
certain focus plane, theoretically speaking, the computation 
should be executed by the following expression: 
2 4 
Ax 2 kx tkaxr^ t kxr. A 
2 4 (8) 
Ay z ky tkyr^ t k,yr^ -^ 
In addition, referred to the creation process of DEM, the 
interpolation method can be selected. 
2. Conclusion 
1) The distortion error calibration method based on 2D 
DLT proposed in this paper, is a method to calibrate 
distortion errors, like the analytical plumb-line 
calibration method by D.C.Brown. This method has no 
relationship with any physical meanings. There are no 
measured distortion error coefficients. Instead, the 
errors on each pixel are measured directly. Those errors 
including various optical distortion errors and other 
kinds. Compared with measured optical distortion 
factors (such as kk; 7... Op; Opa 7...) method, this one 
seems to be more practical. The analytical plumb-line 
calibration method is more suitable for cameras with 
films as the carrier. This method is more suitable for 
digital camera. 
2) This method is fit for CCD cameras with no square 
pixels. 
3) The technical key is to establish accurate plane control 
network and accurately and automatically extract the