er 
the laser 
ditions, the 
"points (all 
| in table 4. 
r the filter 
the ground 
he tile. The 
ize of lying 
sing are as 
  
  
  
  
  
4.3 Removing Noise 
(a) Noise Removing for DSM 
The cross section of a building is shown in Figure 4, and it 
shows results of the Removing Noise Process for DSM. The 
measured raw random points are including the various type of 
noise. In this part, the important things for the Removing Noise 
Process are as follows. 
(1) In the case of DSM, the important things are the surface of 
lying object, and then it is only used to the laser point 
clouds of 1st return. 
(2) In the case of the only 1st return points, the noise points are 
rarely in the inside of buildings. And then, it is necessary to 
perform the Removing Noise Process. 
(3) The noise points of positive direction are not removed, 
because it is including the trees and the lamppost in the 
definition of DSM. 
(4) In this section, the Removing Noise Process performs for 
the noise points of negative direction. 
   
  
  
  
  
  
  
  
180 180 180 
i e M 
160 169. — i] | 160 = t. 
140 140 | | 140 i 
HL 1. | 
120 120 | i 120 
Æ 100 = 100 E 100 
À "a ^ 
= 8 2 dom 
60 | ao 
torent ol mtd Mi lA ERE Mon 
20 j 20 20 
0 Ty à. i dy 
50 i06 ise a 86 i99 iso s ihe 
Horizontal (m} Horizontal [m] Horizontal {m} 
st : 
Raw Data Only 1” Return After Noise 
Figure 4 Cross Section Lines 
(b) Noise Removing for DEM 
The microscopic noise as in Figure 5 is rarely in the DEM after 
filtering process. And then, the DEM is performed with the 
removing local noise process, and the results are shown in 
Figure 6. 
  
Figure 6 after Noise Removing 
5. BUILDING FOOTPRINTS EXTRACTION 
5.1 Subtraction (DSM - DEM) 
The subtraction process (DSM — DEM) is performed between 
the each pixel. Its result is the height of lying object image. 
After the process, it is able to perform in the less influence to 
the height of ground. 
200 
  
Figure 7 Lying Objects Height Image 
5.2 Contour Lines Calculation 
The contour lines are calculated by the height of lying object 
image created in Section 4.4, and it is called “the incipient 
object outlines". The incipient object outlines are including a 
few noises due to miss judgments of the filtering process. 
Therefore, it is able to obtain the outlines of lying object in the 
less influence to the height of ground. At this time, the incipient 
object outlines are including the trees and the vehicles as the 
exception of buildings. 
  
  
  
  
B60 
3 fm] 
Figure 8 Initial Laying Object Footprints 
—223— 
mid