106 
models are composed of many non-linear 
computational elements operating in parallel 
and arranged in patterns reminiscent of 
biological neural nets. Computational elements 
or nodes are connected via weights that 
typically adapt during use to improve perfor 
mance. 
This new techniques can be applied to 
speech recognition, optimization, classification 
of data, especially computer vision and 
pattern recognition. It has several advantages: 
high computations rates provided by massive 
parallelism; a great degree of robustness or 
fault tolerance than Van Neumann sequential 
computers; the significant learning 
mechanism, etc. 
Neural net models are specified by the net 
topology, node characteristics, and training or 
learning rules. These rules specify an initial 
set of weights and indicate how weights 
should be adapted during use to improve 
performance. 
In the society of sun/eying, mapping and 
GIS/LIS, we can identify the following possible 
applications of neural networks: 
- the real-time implementation of computer 
aided surveying; 
- automatic target recognition of aerial 
photographs and remote sensing images; 
- automatic symbol recognition of scanned 
maps; 
- fast implementation on some of GIS 
operations, for example, overlay operation. 
3.4.4 Object oriented grid technology 
From the beginning of automating in 
cartography the discussion between the 
possibilities of grids and vectors played an 
important role. A lot of research time is spent 
on the conversion of one system to the other. 
Although most of the maps that are produced 
automatically are raster files, the large scale 
topographic maps are converted to vector- 
files. Every geographic unit can be 
represented with the help of vectors. The 
vector data can easily be depicted with 
automatic plotters. The technical development 
is oriented on the possibilities of grids and 
vectors. In the past decennia we saw 
sometimes better facilities for vector pictures 
(plotters and screens) and sometimes better 
possibilities for the processing and depicting 
of raster files. Many studies are made for the 
conversion of raster to vectors and from 
vectors to rasters. The vector-raster conver 
sion is relativity simple and can be carried out 
very accurately. The required accuracy is still 
a problem with the raster-vector conversion. It 
is now possible to work completely with raster 
techniques because of the availability of raster 
plotters (raster scanners and raster screens 
were already available). Our research is 
oriented on the only problem that remains, 
the building of object oriented files in raster 
forms. 
3.5 Tite presentation of information 
The research regarding the presentation of 
information plays an important role in our 
Centre for Computer Graphics and Mapping. 
Several topics are: 
- simplification of two-dimensional maps. One 
of the characteristics of a map is the data 
base. E.g. a topographic map on scale 
1:25,000 has a content of about 1Mb 
(vectors). Because the use of a map is 
unpredictable, a user gets superfluous 
information. If somebody is looking to a 
road in the north of the country, he gets 
extra information of the south of the coun 
try. With the computer only the information 
can be presented that is necessary whereas 
the other information is kept in the data 
base. These simple maps are necessary for 
new applications like car navigation; 
- three-dimensional presentation. Although on 
the one hand we see a simplification of the 
two-dimensional maps, on the other hand 
there is a tendency to represent reality more 
realistic with the help of three dimensional 
pictures. Especially with planning, manage 
ment of space and development of simula 
tors, the three dimensional information 
attracts a lot of interest. The most important 
development are the digital terrain models, 
the three dimensional thematic cartography 
and the three dimensional topography; 
- dynamic information. Many processes are 
dynamic, such as changes in weather, traffic 
and transport, pollutions of the environment, 
etc. For these phenomena dynamic maps 
are needed. Much time is spent for this new 
field of cartography.