Full text: XVIIth ISPRS Congress (Part B5)

   
nt 
11 
ie 
Le 
1- 
an Ua Un 
= OO 
o 
Q Q o2 
  
EN 
<-> System/camera calibration 
  
E» Image acquisition 
i 
Detection and extraction of data 
  
  
  
—— Analysis of data 
«t——», Decisions and actions based on analysis 
= P 
  
  
  
  
Self diagnosis 
& 
quality report 
  
fig1 Basic system design 
41 System /Camera Calibration 
The calibration of the cameras and systems are vital 
parts for the system performance. They may be done 
simultaneously in a combined adjustment of the 
camera and system calibration parameters or as 
separated procedures. 
41.1 Camera calibration The calibration of 
electronical cameras have been thoroughly 
investigated (e.g. Bósemann, 1990). The calibration 
must not only take into account the optical system of 
the camera but also the electronic parts. The 
traditional photogrammetric optical calibration 
adjust data to a mathematical model based on 
physical assumptions. This leads to e.g. the familiar 
polynomial equations for the radial distortion. 
Another approach is to calculate the deviations, or 
errors, for each pixel. From the deviations a look-up 
table is created. This method is not concerned with 
the physical background of the errors. It is fast and 
easy to implement and especially suited together with 
the Direct Linear Transformation, DLT. The two 
systems which are not developed by photo- 
grammetrists are using this approach (fig 2). 
  
  
  
  
  
O ms 
@ MacReflex 
e a L1 MapVision 
e O L1 BB TrackEye 
1 1 1 2 
No 
Factory Self 
Calibration Calibration Calibration 
  
  
  
fig2 Camera Calibration 
4.34.2 System calibration The system calibration, 
or the outer orientation, is computed using either the 
DLT or the bundle adjustment (fig 3). The advantages 
  
  
   
  
  
    
   
   
   
   
   
  
    
   
    
    
   
   
   
   
      
    
   
   
   
   
  
   
   
    
    
    
   
   
   
     
    
   
   
    
    
of the DLT is, at least initially, its easy imple- 
mentation and the simplicity of the outer orien- 
tation. The advantages of the bundle adjustment is its 
flexibility in the control, e.g. 1-3D points, distances, 
plumb lines, and its theoretical superiority and error 
propagation capabilities compared to the DLT. For a 
more comprehensive study of the differences 
between the DLT and the bundle approach see 
(Edgardh, 1992). 
  
  
  
  
  
O Ms 
@ MacReflex 
B L] MapVision 
e O r1 Bg TrackEye 
1 1 1 Ba 
DLT Bundle with 
Adjustment self-calib. 
  
  
  
fig3 System Calibration 
Comments The camera and system calibration is a 
part in the measuring process where the differences 
between photogrammetrists and other engineers are 
visible. 
In the camera calibration both of the systems which 
are not developed by photogrammetrists are using a 
factory calibration. This is motivated by the stability 
of the CCD cameras. Both the system measure on 
laser spots or on reflective targets which partly 
reduces the need of re-focusing the cameras or 
changing the aperture, an argument often brought 
against factory calibrations. It may however not only 
be a question of precision but also of reliability as one 
of the photogrammetrists expressed "...never rely on 
a previous calibration. The system should be 
calibrated after the installation and it must be fairly 
easy to re-calibrate both the interior and exterior 
orientation...". 
The choice between DLT or bundle adjustment can be 
difficult in some applications. An advantage using 
the bundle adjustment together with CCD cameras 
compared to analogue cameras is the possibility of 
making several measurements after each other. Two 
systems, #2 @ and #3 Ll utilize this technique to 
calibrate the systems. A known distance is moved 
around in the measuring volume until a satisfactory 
number of observations are made. This is also used 
for the self-calibration of one of the systems, #3 Cl 
This greatly reduces the problem of calibration of 
both the cameras and the system compared to the 
traditional test field calibration. 
4.2 Image acquisition 
In close range applications, the typical image 
acquisition part consists of standard video-rate CCD- 
cameras. A good reason keeping to this standard is 
the large number of fairly cheap electronic 
components, ranging from the CCD-cameras over 
frame grabbers to hard ware implementations of basic 
image processing functions. 
  
	        
Waiting...

Note to user

Dear user,

In response to current developments in the web technology used by the Goobi viewer, the software no longer supports your browser.

Please use one of the following browsers to display this page correctly.

Thank you.