It is increasingly important to deal with updated geographic databases
in order to ascertain any spatial analysis for decision making in urb
an planning. However, if updating a conventional database (e.g., chang
ing the delivery address in a custom file) can be done by means of a S
QL-type language or via a form interface, in the domain of geographic
databases, updating means different things and cannot be performed eas
ily. The essential reason is because updating information is not only
elementary, but rather coming from different sources such as newly mad
e measures, aerial photos, scanned maps, and so on. Indeed, in this ca
se, updating or multisource integration can mean: local cartographic u
pdates - that is to say, to integrate newly made measures into a datab
ase storing old measures: in this case, sometimes some points or some
segments must move a little bit, possibly replacing segments by piece-
wise lines and so on, possibly by means of elastic transformations; gl
obal updates - for instance, using aerial photos in order not only to
detect clandestine buildings but also to automatically shift all point
s of the database; this can be done by defining control points; extens
ion of the coverage - in other words, when we have to fuse maps of two
neighbouring zones whose boundaries do not meet, it is necessary to s
elect control points and to perform rubber-sheeting at the vicinity of
the boundary based on those control points; updating for measurement
or topological error correction - that is, reconstruction of badly str
uctured databases based essentially on topologic information; this can
be achieved on a single layer (e.g., transformation of a spaghetti-li
ke cadaster into a real parcel-oriented cadaster, closure of polygons)
or on multiple layers (e.g., to ensure that a building is not astride
the neighbouring parcels). In this paper we will give an overview of
problems to be solved together with some solutions in order to perform
such different kinds of updating.