CIPA 2003 XIX"' International Symposium, 30 September - 04 October, 2003, Antalya, Turkey 
547 
aiOCK SCANS 
TARGET 
PTSortgimü 
cloud 
<4008 
pt5 mo# 
cloud 
-4*0 
PTS vwnpieo N. cluattr 
cloud /cloud 
X10B3 M 
TsitKGt.es 
T'-COC 
TRIANGLES 
decimated mesh 
SI2S .OBJ 
format 
;J3 
lato dx 10 
6 
7.354 
6.645 
4 494 4 
8.920 
2.141 
81.233 
fondo dx I 3 
5 
2.826 
2.253 
1.042 1 
1.993 
478 
17.959 
parte sx, 7 
6 
7.614 
5 847 
3.535 4 
6.779 
1 627 
61.171 
complet I 
model 
17 
17.794 
14.745 
8.071 9 
17.692 
4246 
160.363 
Figure 8. Cloud point sample. 
A “curvature sample” operation was performed 
(with Raindrop Geomagic Studio software). 
Points that lie in a high-curvature region remain in order to 
maintain the accuracy of the surface curves. In this way it has 
been possible to halve the acquired points 
Figure 9. Clusterization of range map 
(Figure 11), to refine and correct imperfections or errors in the 
surface. The clusters, previously defined has been converted in 
a mesh and then an optimization has been performed. In order 
to preserve surface integrity and details, a method of polygon 
decimation (based on shape preservation), allowing the number of 
triangles to be reduces while maintaining the overall shape of the 
object has been employed. A comparison between the final resulting 
mesh and the original acquired cloud of points has been carried out. 
The deviation is displayed as a colour map (Figure 12). Although 
the original measured data has been reduced by about 15-20%, 
shape description is both satisfying and accurate from a metric 
Figure 11. Example of mesh optimization 
Figure 12. Comparison between the final resulting mesh and 
the original acquired cloud of points. As side color scale 
shows, green indicate deviation <±0.005 m 
Figure 13. Assembled model 
point of view. At the end the entire model has been assembled in a 
reference system and overlapped with photogrammetric restitutions, 
also available for the apse (Figure 13). 
The application of new survey techniques has to be able also to 
produce standard graphic elaborations with these requirements 
consolidated in time; so, simple renderings can be considered a 
base to extract vector data (Figure 14). In this way it is possible 
Figure 10. From point to surface: example of triangulation 
Figure 14. Examples of manual edge extraction