3, MODEL FOR A LARGE-SCALE MAP AND ITS 
REPRESENTATION 
In this chapter we present a model for large-scale maps. 
It is organised in four levels and described in the follow- 
ing for cadastral maps of scale 1 : 1000. In addition to 
three levels (semantic objects, image graph and image) 
that resemble the three levels from computer vision 
(high, mid and low level) we use the graphics and text 
level. It is an intermediate level between the image graph 
and the semantic objects level and takes into account 
that our input data is the result of classification and 
symbolisation by a human. 
Each level consists of objects, operations to be per- 
formed on the objects and relations between the objects. 
In addition there are operations to be performed on and 
relations between objects in different levels. 
For the representation of the model we use a semantic 
network. Our work was inspired by the ERNEST system 
[Niemann et al. 1990]. The nodes of the semantic net- 
work are called "concepts". If a concept is instantiated, 
the result is called an "instance" of that concept. In the 
semantic network we use two types of relations (arcs) 
thatset up our network and connect the nodes: the "part" 
(inverse: part-of) relation and the "generalisation" (in- 
verse: specialisation) relation. A concept built up from 
the part relation consists of other concepts, relations 
between these concepts and operations to be performed 
on these concepts. A concept built up from the special- 
isation relation consists of one other concept and spe- 
cialisation conditions for this concept. 
For the different levels examples of concepts that rep- 
resent objects are given. The operations on and the 
relations between the concepts are explained for the 
levels (see chapter 3.1.) and between concepts in diffe- 
rent levels (see chapter 3.2.). 
3.1. Levels of the model 
  
The first level is called "semantic objects". It contains all 
the objects that are denoted by the map legend. For 
cadastral maps of scale 1: 1000, there are three types of 
semantic objects (see Fig. 1): basic map objects, cada- 
stral objects and topographic objects. 
The basic map objects describe the map (the margin 
built up from the sheet designation, the edition note, the 
scale etc.), delimit the drawing plane (the framework 
and the neatline) and give information about the location 
of the map (the grid with grid coordinates and grid 
intersections). No operations exist but there are rela- 
tions between the locations of the objects. 
The cadastral objects (see Fig. 2) are the parcels, their 
corresponding numbers, the parcel areas, the boundary 
lines and the boundary stones. Only relations exist: the 
cadastral objects build up a layer that is exhaustively 
filled out by non overlapping (ie mutually exclusive) 
parcels. A parcel consists of a parcel area and a corres- 
ponding number. Parcel areas are surrounded by bound- 
ary lines. The boundary lines are related to the boundary 
stones: at every end of a boundary line a boundary stone 
is situated. 
The basic types of topographic objects are: buildings, 
roads (also places), railways, forest, meadows, fields and 
waterbodies (see Fig. 3). 
For the buildings, forests and roads part-of and special- 
isation relations of these objects are given. For the other 
objects the relations are not shown. 
  
Large-Scale Cadastral Map 
Scale 1 : 1000 
me am 
Cadastral Topographic 
Cbiecs hiess 
   
   
   
Margin 
  
   
    
   
     
   
  
Sheet Edition 
Designation Note 
> 
Basic Map Objects 
( Framework ) 
( Neatline  ) 
     
  
  
   
    
  
  
  
Grid 
Intersection 
Grid 
Coordinate 
      
   
Fig. 1: Objects in a large-scale-map 1 : 1000 
(the objects below are parts of the objects above) 
580