ntrol 
id to 
tion) 
d an- 
0 the 
onic 
pro- 
ions 
pro- 
The synchronization and blanking signals are added dur- 
ing signal generation. It was found that some information 
may be lost as the time available during the active image 
period is insufficient to submit the complete image data. 
Furthermore it was found that some pixels on the border 
are significantly darker, assumed to originate form the 
signal mixing. The synchronization signals must be de- 
rived from a master clock to enable a stable locking of 
the frame grabber onto the video signal (Beyer, 1988; 
Baltsavias et al., 1990). 
2.6 The Frame Grabber 
Frame grabbers do often include a multitude of process- 
ing capabilities. The following discussion is limited to 
functional elements involved in image acquisition from 
monochrome imagers. 
Figure 7 shows the major functional elements of a frame 
grabber. The analog front end includes the impedance 
matching circuitry, the DC-restoration, analog offset, and 
analog gain control. Proper impedance matching is re- 
[Analog front end ] —> 
+ i 
Sync detection | — [Databufer — 
Synchronisation 
  
  
Timing generation 
i 
| Signal output 
  
  
  
  
Figure 7 Typical image acquisition components of a 
frame grabber. 
quired to circumvent signal reflections. DC-restoration 
removes the blanking level from the video signal. The lo- 
cation in time of the blanking signal and the quality and 
stability of the subtraction have been investigated. 
Wrong timing of the blanking signal sampling was found 
to lead to a variation of the subtracted signal level of over 
4 grayvalues within up to 100 lines. This can induce a 
geometric displacement similar to local illumination gra- 
dients. The fall-off of the sample-and-hold mecbanism 
used in many DC-restoration circuits was found to result 
in a uniform change of the background which can be dis- 
regarded. This small variation can be seen in Figure 8 as 
a slow increase of the image brightness from left to right. 
The temporal noise characteristics of the frame grabber 
have been shown to be close to the theoretical limit of the 
analog-to-digital converter. More problematic are sys- 
tematic patterns such as those shown in Figure 8. The or- 
igin of both the horizontal bands and the periodic phase 
patterns running at an angle of 45? could not be deter- 
mined. They are assumed to originate from the host com- 
Band with 
lower gray- 
values 
  
Brighter area show- 
ing the “phase pat- 
tern” 
Figure 8 Average frame showing typical patterns of the 
frame grabber. 
puter and other electronic components as they depend on 
the frame grabber board and its position in the chassis. 
The observed patterns can lead to displacements similar 
to those of local illumination gradients. The displace- 
ments from the “phase pattern” are dependent on the size 
and location of the target with respect to the pattern. The 
maximum influence was estimated to reach several hun- 
dredth of the pixel spacing as the peak-to-peak variation 
of the phase pattern is 4 to 6 grayvalues. 
A strong integral non-linearity of the frame grabber was 
detected and appears to be typical for many frame grab- 
bers. It must be considered when performing radiometric 
corrections and radiometric operations but is of no im- 
portance when considering the position of circular tar- 
gets. 
Grayvalue 255 
Figure 9 Image of a vertical line without and with LPF 
showing the assymmetric impulse response 
and ringing when the LPF is used (b). 
       
SR 
ssp sae