consists of 20 targets arranged in a wall and 
two vertical pillars (shown in Fig. 6). The coor- 
dinates of the marked control points are de- 
termined by intersection of directions measured 
by two 2° theodolites located six metres from 
the test field Each control point has 3D-coordi- 
nates with accuracy of 0.3 mm. This control 
field is used for the calibration of CCD camera. 
  
  
  
  
  
  
  
  
Fig.6 A three dimension control field 
4.1.2 Line Jitter Test 
Line jitter is caused by the horizontal line 
synchronization error. In the "plumb-line" test, 
the maximum line jitter error of GOOD LR-1002 
CCD camera has been detected about 0.26 pixel in 
X direction. The RMS (root mean square) is about 
0.08 pixel. In this test we are used four plumb- 
lines, and detected results in Fig. 6. 
| 
Line No. No. 1 | No. 2 | No. 3 | No. 4 
  
  
Max. residual |.261 .260..166 |.243 
  
  
  
  
  
  
  
Min. residual | .079 | .078 | . 070 | . 073 
  
Fig .6 Line Jitter Detected Results 
      
    
    
    
   
    
   
    
   
    
    
       
    
         
   
    
  
    
    
    
    
   
   
    
    
    
  
4.1.3 Calibration of Geometric of CCD Camera 
  
(1) Calibration of the interior and exterior 
orientation parameters. 
The interior and exterior orientation parameters 
can be determined by DLT or spatial resection. 
(2) Calibration of lens radial distortion. 
In order to reduce the lens radial distortion, 
the following correction function is added in 
DLT equation : 
d 
x - (x-x0} + 1? 
dy 
ki 
ki: (y-yO. - r* 
where , r® = (x-x0)%+(y-y0)®, kl is the lens 
radial distotion coefficient. By calculation, kl 
is obtained about 1077, 
"n wu 
4.2 Mark Point Location 
4.2.1 Mark Pattern Design 
In the test selected the mark pattern show Fig. 7 
The test results show , that using pattern (a), 
the accuracy of location is highest . The pattern 
(b for seting it on the moving body is easy 
and it can located with sub-pixel the precision, 
The pattern (c) is made of a small plastic ball 
with diameter 8 mm, pasted with a reflecting paper 
It has strong reflectance under a general illu- 
mination by artificial light . On its image the 
s / n ratio is higher. The system with uncoded 
passive markers allow the analysis of a great 
number of markers. 
(b) (e) 
Fig.7 mark pattern 
  
(a) 
4.2.2 Mark Detecte Algorithms 
The measurement accuracy of the 3d- coordinate 
is depend on detecte accuracy of feature point 
in the image. Therefore center coordinates of the 
mark must be located with the sub-pixel preci- 
sion. The system has one important feature, that 
is the mark detection algorithms, which work on 
the shape and size of the mark rather than on 
their brightness. IT used Multistage Matching-Fit- 
ting Method (Fig. 8), which consisted of following 
algorithms . 
-- Mark Pattern Match