Ultimately, the choice of sensor and technique 
used to generate and correspondingly display video 
images depends on the application for which they 
are intended. 
1.2 Previous work 
The 3-D Imaging Group at Nottingham (formerly 
Trent) Polytechnic has been involved in the 
development and application of three-dimensional 
visual techniques for a number of 
years$, /,8,9,1 0, Initial work included the 
construction of a stereo camera system, mounted 
on a teleoperated bomb disposal vehicle!!. As a 
result of this work, it became apparent that the 
three-dimensional position of an object could be 
determined by the use of such a system! 2. 
At first this involved the introduction of measuring 
marks into the video signal and necessitated an 
operator to position the marks manually over the 
points of interest. These marks were then used to 
calculate the position of the object. The success of 
this technique prompted research into the use of 
such camera systems in robotic guidance 
applications. This has involved controlling a robot 
arm from information obtained from a 
stereo-camera system13. This arrangement has 
enabled the 
determined, and the robotic manipulator to be 
position of an. object to be 
moved such that this object can be grasped. This 
was done without prior knowledge of the position 
of the object. 
Concurrently, research has involved the production 
of a three-dimensional X-ray machine based on 
linear array sensors! 4. This machine is at present 
being used for work concerning the extraction of 
three-dimensional co-ordinate information from 
X-ray images! 5. 
1.3 Research Aim 
. The television type sensor, used in the majority of 
machine vision applications, was primarily 
designed to produce images that were suitable for 
humans to observe. However, these images may be 
far from ideal when the task of visual analysis is 
to be undertaken by a robot! ©. Hence, it is an 
objective of this research to demonstrate that 
there exists alternative methods of producing both 
two- and three-dimensional visual systems. 
Initial research at Nottingham has demonstrated! ' 
that non-standard television type sensors, in this 
instance the line-scan camera, can be used to 
produce vision systems. Such systems have been 
calibrated and have subsequently returned a 
sub-millimetre degree of accuracy in all three 
co-ordinate axes. The current aim of the research 
leads on from this initial work and involves 
investigating the line-scan device further. 
This paper gives brief details of the research 
conducted using the line-scan device to produce 
stereoscopic images and suggests possible uses for 
such systems. 
2. BACKGROUND 
2.1 Two-Dimensional Imaging Principles 
2.1.1 Image Production The line-scan camera 
consists of a single line (or column) of contiguous 
picture elements or pixels (Fig. 2). It is often 
referred to as a one-dimensional array and the 
number of pixels in a typical device can range 
from less then 256 to over 6000. The principle of 
18,19 is similar to the 
operation of these devices 
two-dimensional CCD sensors used in most 
modern television type cameras (Fig. 1). 
PICTURE ELEMENTS 
TELEVISION TYPE SENSOR 
  
Figure 1 Photosensitive Area of Television Type 
Sensor