grammetric system with Megaplus 4.2 cameras allows 
to determine each two seconds the coordinates of about 
50 object points (all points at the same time). 
To compare the different systems by means of the 
same moving object a rotation bar commonly applied to 
calibrate the SMART laser tracker was chosen (Fig. 1). 
The motor driven bar rotates approximately 10 times 
per minute at most. This maximum speed was used for 
all investigations. A target fixed to the bar moves on a 
circular track during rotation. The maximum radius to be 
performed is about 550 mm according to the length of 
the bar. The accuracy of the circle which corresponds to 
the track of the attached target amounts to less than 
5 um, i.e. deviations from the nominal radius and the 
nominal plane in which the circle lies are expected to be 
less than 5 um. 
The motion measurement systems use different targets 
fixed to the bar at slightly different positions. Thus, the 
radius generated by a moving target is not appropriate 
to compare the measurement results. A best fit circle 
was calculated by least squares adjustment from the 
3-D coordinates of all points of the track. The root mean 
square (rms) of the deviations of these points in relation 
to the best fit circle and best fit plane serves to estimate 
the quality of the measurement system. 
3. ON-LINE 3-D COORDINATE MEASUREMENT 
WITH V-STARS/M 
The V-STARS digital close range measurement system 
for industrial photogrammetry is able to determine 
position and form of spatial objects from digital images 
(Brown and Dold, 1995). On-line 3-D coordinate 
measurement can be performed with digital Kodak 
Megaplus 4.2 cameras providing a resolution of about 
2000 x 2000 pixels (Fig. 2). Due to the fast throughput 
of these cameras two images are imported into the 
computer memory within a second and 3-D coordinates 
of about 50 object points are computed in one 
additional second. Thus, object tracking can be carried 
out in a cycle of two seconds. 
The camera setup to record the rotation bar is shown in 
Fig. 3. Five targets were fixed to the bar which was 
located in front of a wall with a cluster of 20 stable 
control points. Within two seconds, V-STARS/M 
acquired the images from the two Megaplus cameras, 
computed the coordinates of the five targets and utilized 
the control points on the wall to update the position and 
orientation of the cameras. Thus, the accuracy of the 
measurement process could not be influenced by vibra- 
tions of the camera locations. Because one full rotation 
of the bar takes six seconds the bar had to rotate 
several times to obtain a number of different positions 
on the circular track. A series of measurements were 
done with 8 records per track at minimum up to 145 per 
track. The centers of gravity of the five targets per 
record were used to calculate the best fit circle. 
  
Figure 1 Rotation bar used as moving object 
  
Figure 2 V-STARS/M system 
Wall including 20 control points 
  
8 Rotation arm 
L5m | 19m 
: E. 1.9 m 
2. 5m 
eG @ cameras 
Figure 3 Camera setup 
472 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B5. Vienna 1996 
  
  
Devia 
are g 
show: 
the nt 
the c 
devia 
40 un 
If an : 
LED 
objec 
came 
light 
meas! 
using