36 
with Stuttgart University, developed a system on 
the Planicomp C 100 analytical plotter (Giilch, 
1984; Pertl, 1984), whilst Kern of Switzerland 
introduced the Kern Correlator, based on the 
DSR11 (Bethel, 1986). 
The third major European manufacturer, Wild 
Heerbrugg of Switzerland, did not attempt similar 
efforts using any of their range of analytical 
instruments. However, outside of the commercial 
field, some work has been done at ETH Zürich on a 
modified AC1, also mainly with a view to 
automatic DTM generation (Baltsavias, 1988). The 
experience gained with this instrument provided a 
basis for much of the work done on the S9AP. 
2.2 Aspects of the S9AP Construction 
The S9AP is an integral part of System-9, the 
geographic information system (GIS) launched by 
Wild Heerbrugg in 1985. It is well documented in 
the literature (eg Schneeberger and Bürgermeister, 
1987), but some aspects of the hardware 
construction must be outlined as they have 
implications on the work of the project. 
The viewing system of the instrument is of 
particular interest. The design of the S9AP was 
based on the existing BC range of instruments, 
rather than the AC1 which was known to be of 
slightly higher accuracy. The essential difference 
between the two is the means of image 
measurement: in the AC1, the entire optical train 
remains fixed whilst the photo carriage moves in 
two directions; in the BC range and in the S9AP, 
the carriage moves only in the Y direction, whilst 
measurement in the X direction is effected by 
translation within the optical train. The major 
advantage of this is a reduction in the size of the 
instrument, at the cost of a small reduction in 
accuracy in the case of measurements by manual 
operator. The consequences for digital 
measurements are somewhat more significant, 
however, as will later be described. 
Plate positioning is done via a feedback loop using 
servo motors and linear encoders of 1 pm 
resolution. Instrument calibration relates the linear 
encoder scales to a stage coordinate system defined 
by a grid of crosses engraved on each photo 
carriage, the relative positions of which are 
determined to high accuracy in the factory before 
installation. The quality of this calibration can be 
affected by the optical design as described above. 
2.3 The S9AP with CCD Cameras 
The S9AP at ETH Zürich has a modification 
allowing the attachment of most types of CCD 
camera. The optical and structural design and 
implementation of this modification was carried 
out by Wild engineers in Heerbrugg, with the 
consequence that the instrument in its manual use 
loses nothing in terms of optical quality. The 
important features of this realisation are: 
• The cameras are mounted externally at the rear 
of the instrument via a C-mount fitting. Any 
camera using such a standard thread can be 
simply “screwed on”. 
• The magnification is approximately unity. It can 
be changed by a translation of the objective lens, 
but only within a small range. The lens is, 
however, not easily accessible. 
• The distance of the camera from the stage plate 
is rather long, having the disadvantage that 
systematic effects arising from the opto 
mechanical design are somewhat magnified. 
On the other hand, the long focal length of the 
objective lens (~ 30 cm) means that any effects 
of distortion are minimised. 
Figure 1 shows the relevant part of the S9AP optics 
and the positioning of the camera (one side only). 
Figure 1. Part of the S9AP optical train showing 
positioning of the CCD camera mount. 
The camera used for the investigations was the 
Sony XC-77CE, an inter-line transfer device with 
pixel size 13.6 x 11.1 pm on the stage, giving an 
image size of 8.0 x 6.4 mm. 
The image acquisition hardware used comes from 
the Maxvideo family of products manufactured by 
Datacube Inc in the USA. The boards operate on 
the VMEbus, thereby allowing all system 
components - host computer, S9AP and image 
acquisition hardware - to be controlled as a single 
system from one workstation. 
3. System Calibration 
3.1 Purpose of Calibration 
The primary purpose of calibration in the context 
of an analytical plotter fitted with CCD cameras is 
to derive a relationship between the acquired 
digital images and the analogue imagery being 
digitised. Once this relationship is determined, it is 
also necessary to know about its accuracy and 
stability. Both are influenced by many factors, and 
hence the whole procedure encompasses many