Roland Billen 
  
It is the most common way. The measurement provides the 3D geometry. It could be done with classical survey 
material, GPS, photogrammetry, laserometry, radar interferometry. 
Estimation 
This mode is based on the estimation of the third dimension from the 2D footprint and from a specific information like 
the density, the number of roof and their height, the context, etc. 
Generalisation 
The 3D geometry can be found by some generalisation techniques. These techniques must be fixed for each object and 
be part of the metadata. 
Desaggregation 
The 3D geometry of a lower object can be derived from the geometry of a upper object. In this case also, the techniques 
used must be fixed and become a metadata. 
This conceptual subdivision is not useless even if the user needs are well known. For the object attributes, similar 
reasoning must be done. This object approach is then directly linked with the data acquisition. The link between 
conceptual model and 3D processing is not yet established. 
5 PRIMARY APPROACH OF A GLOBAL 3D URBAN MODEL 
Actually, 2D reference urban database are nearly exclusively produced by public authorities. Their goal is often to 
provide information to the maximum number of users. Because of the high cost of the 3D data, this kind of solution, i.e. 
production by public authorities is a good solution for a lot of users. The problem lies in the nature of the 3D 
information. To be useful for a large number of users, the 3D model must contain a common information base like in 
2D databases. Thus, a global 3D urban model must present different subdivisions. This 3D subdivisions should 
correspond to the 2D subdivision usually used by public producers. This approach meets our will of global spatial 
thinking and then is not linked to a specific application. It is build on the generalisation of the actual 2D urban 
subdivision and of 3D city information (already) collected. The following figures are only schemes. They are not 
designed with a specific formalism and their interpretation is not strict. According to OEEPE study (Fuchs, 1998), 3D 
city information model should include : - buildings; - vegetation; - traffic networks; - public utilities; - 
telecommunication; - others. These are linked with potential urban objects. 
  
  
Building Vegetation 
Ensemble Ensemble 
? 
Others Building Vegetation Traffic networks Public Utilities 
Cell 
        
      
   
Administrative 
object 
Census Track 
object 
Postal 
object 
   
Figure 3. Starting point of the categorisation of 3D city information 
In figure 3, 3 objects were added : - the cell; - the building ensemble; - the vegetation ensemble. The cell is introduced 
because it corresponds to the basic human division of his built environment. It could be a level of a building, above or 
below the ground. Cell could be composed by n-ary cells corresponding to more specific criteria (architecture, social, 
etc.). A building is composed at least by one cell. The building ensemble is a set of building grouped by neighbourhood 
and / or similarity criterions. The vegetation ensemble is a set of vegetation unit also grouped by neighbourhood and / or 
similarity criterions. Telecommunication objects such as antenna or call box can be basic elements but products such as 
digital surface model are in fact objects derived from other urban objects. Administrative objects, cadastral, postal and 
census track are other potential 3D objects. These are 3D extension of the classical 2D subdivisions of urban space 
(Rouet, 1993) 
  
82 International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B3. Amsterdam 2000.