International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B6. Istanbul 2004
5.1 Functions resident in the software platform
The ArcGIS software platform used in this work contains some
functions dedicated to generalization. The main functions used
in the course of this work are the following:
- the function Create Centerline
- the function Simplify Buildings
- the function Simplify Lines
These three functions accessible in the ArcToolBox
program. The Create Centerline function was applied to the
covers containing the road and fluvial networks on the entities
that went into the definition of the single road shoulder or river
bank.
The Simplify Buildings function simplifies buildings and other
prevalently squared-off entities, thus reducing the details of
their perimeters and maintaining the essential features of their
shape and original dimensions. Buildings are, in general,
orthogonal areas and this kind of simplification maintains and
increases their orthogonal nature.
This function was applied, in the course of our experiment, to
covers containing the building and to those containing the other
buildings represented with areal entities.
The Simplify: Lines function was not used in the generalization
process. It was in fact possible to use the Generalize method,
which implemented the Douglas-Peucker algoritm, for the
Polyline class.
We then implemented a procedure dedicated to simplification of
linear elements which exploited this possibility within the
development environment.
are
5.2 Procedures implemented for the first time
The procedures that were implemented ex novo can be classified
on the basis of the kind of operation they perform:
Procedures for restoration of congruousness and continuity
e Procedure for the merging of lines
e Procedure for joining broken lines
Procedures for logical thinning out
e Procedure for thinning out punctual objects
e Procedure for thinning out lines
Procedure for thinning out areas
Procedures for simplification of shapes
e Procedure for line simplification
Procedure for the simplification of areas
Procedures for enlarging areas
e Procedure for enlarging buildings
e Procedure for widening buildings
Procedures for area aggregation
e Procedure for aggregation of buildings
e Procedure for aggregation of areas of soil occupation
General procedures
e Procedure for generalization of buildings
Procedures for transformation of the original and changes in
attributes
Procedure for the collapse from areal to punctual entity
Procedure for change of the cartographic code of the entities
For brevity's sake, attention will be focused on the procedure
for generalization of buildings.
74
6. PROCEDURE FOR GENERALIZATION OF
BUILDINGS
The aim of the procedure is to increase the dimensions of the
entity up to the predetermined minimum values while
maintaining the original shape and at the same time avoiding
superimpositions and ensuring respect for the relative minimum
distances through elimination and aggregation.
For generalization of buildings the procedures of enlargement,
thinning out and aggregation are used jointly and are
implemented within the Generalize buildings function. which
can be implemented in the ArcMap environment.
The first procedure applied is that of enlargement, which
increases the dimensions of the entity up to the required
minimum. Enlargement is applied progressively following a
number of steps determined beforehand. At each step any
superimposition conflicts that may arise are indicated. Once the
conflict has been highlighted, a special procedure chooses the
next operation to be applied (thinning out or aggregation) on the
basis of a cost function employing the relationship between the
dimensions of the two entities as a variable. At the end of the
enlargement procedure any proximity conflicts are pointed out
and also in this case they are addressed gradually by varying the
minimum distance up to the final value. Both the aggregation
and thinning out procedures lead to a resolution of conflicts and
respect for the no superimposition (topological) and separation
distance (proximity) constraints. Depending on the typology of
the two objects and the density of the conflicting buildings, one
or the other of the two procedures is found to be the more
suitable.
One way of making the choice is to compare the dimensions of
the two buildings. If they are more or less the same size the
ideal solution is that of aggregation. But if one is clearly larger
than the other the best solution is to eliminate the smaller one.
Another method for optimizing the choice is to evaluate the
building's density. Indeed, while aggregation does not make
room for the other entities, thinning out produces too much
open space. An optimal balance between aggregation and
elimination leads to a result that is more harmonious and
respondent to the reality it is representative of. The subdivision
of the process into a defined number of steps allows simulation
of a continuous enlargement, with the highlighting of conflicts
as soon as they occur. This ensures that conflicts occur
prevalently between a limited number of entities and can be
resolved easily two by two. Application of enlargement in a
single step would lead to a large number of conflicts, thus
complicating the choice of the entities to eliminate or aggregate.
In the same way, in the case of proximity conflicts, this way of
proceeding allows dealing first with the entities closest together
and then all the others. The more the steps in the procedure the
better the final result will be; however, this creates an excessive
process cost both in terms of time and memory. It has been
found that three to five steps are sufficient to produce a good
result. With more than five steps the increase in process cost
does not produce a corresponding improvement in the results.