International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B6. Istanbul 2004 
  
superimposed or are too close together, thus generating a 
conflict. Resolution of such conflicts can come about in three 
different ways: by eliminating one of the two, by aggregating 
the two or by shifting one or both. 
Thinning out solves such conflicts through elimination of some 
entities. 
6.3 Procedure for thinning out buildings 
This type of thinning out is performed for buildings by a 
procedure inside the Generalize Buildings function which can 
be activated by means of a pushbutton. In this case the thinning 
out of areas is performed on the basis of considerations of a 
contextual nature together with their aggregation by the 
Buildings Generalization function. In any case, the function can 
be programmed to perform only the thinning out operation if 
aggregation is not desired. 
The need to perform a reduction in the number of buildings 
arises every time the latter become superimposed owing to 
enlargement operations, or when the minimum distance 
between them is not respected and visually they can no longer 
be distinguished one from the other. 
It is then necessary to define a minimum-distance-between- 
buildings parameter, below which thinning out is performed; it 
is also indispensable to define a criterion for the choice of the 
entities to maintain. The minimum distance value can be 
assumed as equal to the minimum value allowing a visual 
distinction between entities. Lacking semantic information, the 
choice of entities to keep and those to eliminate is made by 
attributing greater importance to the larger one and sacrificing 
the smaller one. 
Since the need to thin out is a consequence of a reduction in 
map space caused by building enlargement, this procedure must 
be activated after, or better still, at the same time as the 
enlargement procedure. 
The procedure operates as follows: 
e it examines all entities in the area and identifies those 
which, following enlargement, are superimposed (in 
first analysis) and those which are less than the 
minimum distance apart (in second analysis); 
® when a conflict is revealed, the algorithm proceeds to 
the elimination of the entity having a smaller surface 
area. 
The procedure can be modified to perform thinning out only on 
the basis of proximity and thus without enlargement of the 
entity or on the sole basis of superimpositions that occur. If so 
desired, the enlargement performed to reveal superimpositions 
can be eliminated, thus returning to the original representation 
of the entities on which only thinning out was performed. 
The algorithm naturally contains some limits: it does not allow 
optimization of the thinning out to minimize the number of 
buildings eliminated to comply with the constraints imposed. 
The result depends on the sequence in which the entities are 
scanned and can therefore lead to different results depending on 
the order in which they are taken into consideration. 
However, it does have the advantage of being easy to 
implement, of having a high degree of efficiency, of being 
governed by a single parameter (minimum distance) and/or by a 
single topographic consideration (superimposition) and of 
resolving each conflict as soon as it occurs and immediately 
choosing the entity to be eliminated on the basis of a single 
comparison (surface area). 
Below, we present an example showing the joint application of 
thinning out and enlargement in a built-up area. 
76 
ne 
» Un d 
i Pe Pd & 
ms Plo, BRL» De 
La TL 
ET 
Fa n «t Am "eg N P" 
5 nh, Tg, "lag Di e ^ y 
a, 8 "m Cay pg preted ar B? 322 a^ 83 
» »a-a - 31] 4 2, Self pn 
» J #7 SE RES , Ly pf Ney al an % 
g^ > vo, PITRE IR AS NZI RG 
a.“ w 3 a vy fé, tar 
teg ze a. Ley AN 
wi Bien Ba dan sw) Ne = INCH 4 ^ AD IH 
a *. > Wee an" "anta n ord D RTE) 
NUR PCR o ETT 
: * = "gy » * " 
. a et? "3 ry P REMIT 
Se «9*9. 9 v DA Va Ry p 
XO RM, Ct v 1 By v A 
5 42 wt 
"RE S 
x Were, 
SiS... 
- ~~ = 
Figure 5. 
x 
Y x v 
w C 8 
€" 1 ant a 
"a EDU ^s * 
4 u = 
EN Mta, x Ard 
as uo 5 SO tw Mare JAS 
S gs aut 4 15 4M 4 y, 
g" $9 et = BY Rt e dn 
* » * = 9 1 B wt & br [ l a 
s Jy ate gn uk atl 
=... st + . à 
» e Be v : 1 a t 
* 
£ AO Jin 
hy? 
um u 
6.4 Procedure for aggregation of buildings 
For buildings, aggregation is implemented both in the case of 
superimposition and proximity conflicts. 
Here too the aggregation operation can take place only if the 
entities are characterized by the same map code but, since the 
operation is activated inside a general function dedicated to 
building generalization, this is performed on the basis of 
considerations of a contextual nature together with their 
thinning out. In any case, the function can be programmed to 
perform only the aggregation operation if elimination is not 
desired. 
The procedure operates as follows: 
e it examines all entities in the areca and identifies those 
which, following enlargement, are superimposed (in 
first analysis) and those which arc less than the 
minimum distance apart (in second analysis); 
e once a conflict is revealed the algorithm proceeds to 
aggregate of entities; 
€ a buffer of predetermined size is applied to the entities 
to be aggregated. This operation is performed to 
obtain a kind of simplification in shape of the 
resulting entity; 
€ (he values of the attributes of the resulting entity are 
modified. In particular, the "area" field is updated by 
assigning to it the sum of the values of the aggregated 
entities; 
€ the resulting entity is scaled to bring its surface arez 
to the value indicated in its area field. If in fact two 
entities have to be enlarged in order to be correctly 
Int 
Pr 
on 
“Th 
all 
ori 
lev 
ob 
gel 
dat 
pat 
fin 
em 
pic 
stri 
an 
SCC 
an 
vel 
ap] 
Ru 
Pr 
Pu 
Fal 
GI: 
Sal 
AS 
ES 
wil 
of. 
Fal 
ST 
(2) 
Pet 
tec 
We 
Ot! 
C