5.3 Rough Building Footprints Extraction 
In this section, the incipient object outlines are analyzed for 
shape. And then, the analysis is classified to the buildings and 
the others. The flowchart of building discrimination process is 
shown in Figure 9. 
  
  
5 2 b) Extraction wih e 
Compbxty of Polygons dE 
i oer 
Pe 
| Contour Poygons 
| 
| 8 i 
afrerNose Removing 
SZ) E RES 
- as 
  
N 
T= (_ Polygons J 
Yes ET 
Rum 
| 
| Prospect of 
(_ BuitthgFootprnts / 
Ss Bug FO z 
TT 
be A “00 + 
Kl a a a a a Sn Zn m an un Zr Sn a Sn SS Sn Sn Sr (rn See Fee Seen m en en en Ama m Sn “sem an Zen ar a m a m“ | 
52 () Extraction wih WE FATUM o un 
SiniBrty of Polygons Cakuhtbn of Polygons 
Y | CenterPosrtbn & Ara 
  
  
  
  
  
  
  
  
uS 
E c reb Lua. 
| Center Ponts | NS — —-— ar 
| ofPoygons | ee | 
s d | Distance among center] | De SE | 
| Y | Polgons 
Jo n : | Polis of Pons, (o On DS d 
| TN Cieatbn fiom 
| CenterPonts of Poygons 
Grouphg Process | 
Distance among Center Ponts « 5m 
& Ama RatD < 10% 
  
Kondirpn for Same Goup] 
  
  
== 1 
Extraction of Rough Builihg Footprints ; ZA 
Sekctedby M ı Permeter in each group | Grouphg | 
[ rox J 
  
  
  
  
  
Figure 9 Flowchart of Initial Building Footprints Extraction 
(a) Discrimination based on Area of Polygons 
The area of polygons that is extremely large and extremely 
small is excluded as the other of buildings. The characteristics 
of objects are shown in Table 5. 
  
Polygon Area is | Ÿ Influence from the ground due to filtering 
extremely large error 
Y The plural neighbouring buildings are 
combined 
  
Polygon Area is | v' Influence due to trees, vehicles, lamppost 
extremely small and the others 
v Influence of measurement noise 
  
  
  
  
Table 5 Discrimination based on Area of Polygons 
(b) Discrimination based on Complexity of Polygons 
In this section, the incipient object outlines are discriminated 
into prospects of building footprints or the others according to 
complexity of the incipient object outlines. The formula of the 
footprints complexity is as follows. And the characteristics of 
contour lines are shown in Table 6. 
Perimeter of Polygon (1) 
Complexity = 
Area of Polygon 
  
Small Complexity | v Selected as prospect of buildings footprints 
  
Large Complexity | v' Excluded as trees or spike noise 
v Excluded as combined neighbouring 
buildings 
v' Excluded as influence of the ground based 
  
  
  
on incomplete filtering process 
  
Table 6 Discrimination based on Complexity of Polygons 
The results of discrimination based on the area and complexity 
are shown in Figure 10. The excluded polygons are shown by 
the dotted line. 
  
Xni 
  
  
  
gr 
590: 
Xin] 
Figure 10 Removing Needless Polygons based on Area and Complexity 
(c) Classification based on similarity of polygons 
Seeing results in section 5.3(b), the building contour lines 
obtained as the prospect of building footprints are almost placed 
on the same position and have resemble concentric shapes. The 
width of similar contour lines is approximately 2m, it agrees 
with the sampling interval of laser point clouds. The prospect of 
building footprints is grouped based on similarity of contour 
lines. And then, the incipient building footprints are extracted 
from the each group. 
As the grouping process, the distances among central points of 
polygons are used. At this time, the TIN created from the 
central points of polygons, and the line segment length of TIN 
are calculated. The similarity of contour lines are discriminated 
to the same group due to the line segment length and the ratio 
of polygon area is less than threshold. The incipient building 
footprints are extracted due to the shortest perimeter in the same 
group. In this case, the extracted building footprints are selected 
with relatively high position. Then, it is able to reduce the 
influence of building shade. And, the extracted building 
footprints have the merits that are less microscopic uneven as 
the most stable shape. 
-224- 
  
  
x N dc). 
A