m and 
ematic 
they 
ot of 
er and 
on to 
It is, 
e and 
zrated 
s give 
ce of 
are 
e and 
the 
ded. 
| and 
using 
and 
s the 
ngle, 
into 
nding 
rical 
n in 
s no 
ndard 
must 
faces 
oles. 
» but 
e.g. 
Thematics 
  
Geometry Co 
Topology Metrics y 
Fig. 1: Geometry and Thematics 
2. DATA CAPTURE 
Since geometry (topology and metrical information) 
as well as thematic information have to be 
captured, their complexity bears an influence on 
the method of data acquisition. 
2.1. Current Photogrammetrical Data Capture 
  
Unfortunately, most current applications of 
photogrammetry are connected with graphic oriented 
mapping systems. These standard photogrammetrical 
software only records line and point data without 
classifying topological relationships and thematic 
groups, i.e. it only provides a copy of the map in 
digital form. As a consequence of the weaknesses 
of these data models used hitherto, much useful 
information were not recorded. 
Working with a photogrammetrical data capture on a 
feature-oriented but off-line connected 
workstation, a lot of geometrical information are 
unfortunately lost in the translation process 
between the analytical plotter and the GIS 
database. 
2.2. Request to Data Acquisition Procedures 
  
Methods of data acquisition must: 
- support geometric and thematic modelling 
- process complex spatial, geometrical objects 
- apply complex rules of consistency 
- permit relationships between objects 
- be simple and quick to use and have a good 
cost/benefit ratio 
- facilitate data acquisition over large areas 
Two different procedures have to be considered with 
respect to data acquisition: 
1. acquisition of basic data 
During basic data capture the operator has to 
record a great number of geometric and 
thematic data vithin a short time. He wants 
to see immediately the status of his work. 
2. up-dating spatial data bases 
During up-dating, the operator records new 
data selectively. Therefore he needs to 
compare the old data status with the new one. 
With both methods the user wants to make on-line 
consistency checks to establish a consistent 
database in order to reduce subsequent corrections. 
However the verification procedures are time 
consuming and analytical photogrammetric 
restitution systems are expensive. 
11 
3. DATA CAPTURE STRATEGIES 
Two data capture strategies have to be 
differentiated: 
— the unstructured data capture 
- the structured data capture 
Ultimately, the data at the outputs of all the data 
acquisition systems conforms to the data base 
model. The final product does not therefore 
represent the difference between them, but rather 
the way in which it is produced, and this depends 
on the volume and complexity of the data. 
3.1. Structured Data Capture 
  
Every element or object is entered in a single 
operation together with all its topological and 
thematic characteristics, while at the same time 
observing all the relevant conditions, 
consistencies and parameters. 
Even at the digitization stage, it is of advantage 
to distinguish between the geometrical symbol, line 
and region elements, i.e. data acquisition should 
be structured: 
- when "only few new" data have to be added to a 
(much larger) block of old data, because data 
is accessed selectively during up-dating. 
- when the data is of a complex nature (e.g. 
each element belongs to a different thematic 
group and has different attributes). 
- when a number of regulations and special 
conditions require continuous interactive 
intervention on the part of the user (e.g. 
general surveying). 
3.2. Unstructured Data Capture 
  
During the unstructured data capture the operator 
can digitize a large amount of data, called 
"spaghetti, quickly and efficiently. He only has 
to assign one (or several) thematic codes to each 
spaghetti and he is thus free from having to do 
structuring or description work. 
He can digitize a large amount of data, quickly and 
efficiently, which is topologically structured and 
prepared for entry into the information system in a 
further work stage. 
Functions such as "snap to line" or "snap to point" 
are available when digitizing spaghetti. Extensive 
elements do not have to be closed. 
Unstructured data acquisition is to be preferred: 
- when a "large volume of new" data has to be 
recorded, e.g. when recording data for the 
first time. It is of advantage not to have to 
pay attention to data structure or topology 
when recording basic data. Complicated 
consistency checks can be carried out later. 
- when the structure of the data is 
uncomplicated, i.e. when there are only few 
(or perhaps only one) thematic groups and all 
the attributes are the same. 
- when data acquisition can be performed 
automatically and structuring for the most 
part requires no intervention by the user. 
As a solution we propose an on-line data capture, 
which takes place in three steps, and the extension 
of the photogrammetic restitution system with an 
image superimposition system.