3. PRESENT RESEARCH 
3.1 Linear Displacement Vision System 
Initial research at Nottingham Polytechnic has 
involved producing stereoscopic line-scan images 
from linear motion of the camera system relative 
to the object of interest and vice versa. A diagram 
of the system used is shown in Figure 14. 
  
  
  
  
  
  
> 
À N DIRECTION 
OBJECT OF /i i\ OF OBJECT MOTION 
INTEREST / [X — 
um 77 + I A 
L Ld T Y 
| 7 | i s 
pe 3 + A + a 
7 ; i A 
f 3 Eu 
Fi) b. 1 
f ,"LOBMERO WX. À 
/ "FIELD OF "« 1 
ls VIEN SUN 
Let V 
P > 
^ 1l >» 
/ \ > 
/ \ 
LINERR / \ 
RRRRY ; X 
CAMERA | ||/ x 
Figure 14 
3.1.1 Experimental System The experimental 
arrangement consisted of:— 
1. two 1024 element line-scan cameras; 
2. a variable camera positional controller; 
3. two 1024 pixel by 1024 line by 8 bits 
deep framestore; 
4. an object translation stage. 
The operation of the experimental system is 
described below. 
With reference to Figure 14, an object is placed 
on the translation stage. The cameras separation 
and convergence is adjusted so that the field of 
view of each overlaps, and provides a stereo 
region within which the translation stage is to 
operate. The scan speed of the cameras is 
adjusted manually, along with the translation speed 
and the start point of image capture. The 
framestore is initialised and the relevant software 
  
     
executed to allow simultaneous image capture of 
the object as it passes through the stereo region. 
The object is subsequently moved in front of the 
cameras. 
At a software controlled point in the movement, a 
hardware signal from the object motion controller 
starts image capture. When a full two-dimensional 
image from each camera has been generated, 
object motion is stopped at the opposite end of the 
translation stage. 
This initial run allows certain parameters within 
the confines of the system to be monitored and 
adjusted accordingly, and therefore allow the most 
suitable images to be obtained of the object. 
These parameters include the focus for each 
camera, the image capture start point, the extent 
of the field of view (does the interaction of scan 
and motion speed achieve the required field of 
view?), and the necessary amount of object 
illumination. With these parameters adjusted to 
the optimum setting, images can be generated of a 
calibration frame and these used to calibrate the 
stereoscopic arrangement. Objects of arbitrary size 
and shape can then be placed on the translation 
stage and perspective views produced of them. 
These 
three-dimensional co-ordinate information from 
images are used to obtain 
the object space. 
3.1.2 Results A series of experiments has been 
conducted with this arrangement of line-scan 
17 
cameras 
. This work has demonstrated that such 
an arrangement of these devices can be used to 
resolve three-dimensional co-ordinate information 
from an object workspace. Table 1 below provides 
a summary of the results obtained from this work.