International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B2. Istanbul 2004 
  
  
  
©) Orientation Scene representation 3-d (x,y,z) 
i GPS/INS m o d Object model )(Surface reflectance) 
© Z À J v 
Sampling 2-d Sampling | 2-d Sampling 
High-resolution High-resolution 
2:8 Gy) range pue reflectance image 
@ Laser pulse dics 
Temporal 
signal form 
Convolution | 
A 
Scanning 
  
  
  
  
  
  
        
   
  
  
     
    
   
  
   
  
High-resolution 
intensity cube 
High-resolution 
intensity image 
@ Pa Y Y 
    
      
  
ZOm-Hupnumzmao >= >00 
  
  
  
| Spatial undersampling [ 
Receiver i 
Intensity cube 
  
  
  
  
A Region edges 
NHL D>ZP pDADO 
  
d Parametric boundary function 2) 
i reel 
Edge position & Neighboring edge 
orientation estimation|7 following algorithm 
(x;y,z,d,8) 
    
   
  
  
  
  
  
  
  
    
  
  
  
  
Segmentation 
  
  
3-d Region based 
segmentation 
(LR) | x X A 
Pulse property extraction 
E Pulse C és Y 
power f peaks Position 
X 3 
gt A 
Time-range 
conversion J 
Travel time 
measurement 
WEN À 
Detection filter B 
(Local maximum) E 
   
  
  
  
    
Temporal 
signal anal 
       
(P,x,y,z,M) 
  
  
   
  
  
Signal preprocessing 
Coney snd 
       
  
  
  
    
Signal with 
improved SNR 
Matched filter (= 
(Correlation) 7 
|: -] 3-d Gy, 
  
(Simulation) 
3-d GGy, : 
© Intensity cube ! 
Measurement (Measurement) 
3-d (x,y,t) N 
Figure 2. Simulation setup: data generation (left), and analysis (right) 
Beside the first or last pulse exploitation the complete 
waveform in between might be of interest, because it includes 
the backscattering characteristic of the illuminated field. 
Investigations on analyzing the waveform were done to explore 
the vegetation concerning the bio mass, foliage or density (e.g. 
trees, bushes, and ground). NASA has developed a prototype of 
a Laser Vegetation Imaging Sensor (LVIS) for recording the 
waveform to determine the vertical density profiles in forests 
(Blair et al., 1999). The spaceborne Geoscience Laser Altimeter 
System (GLAS) determines distance to the Earth’s surface, and 
a profile of the vertical distribution of clouds and aerosols. In 
some applications (e.g. observation of climate changes), clouds 
are objects of interest. But, clouds can also be considered as 
obstacles, which limit the visibility of the illuminated object. 
Apart from the range measurement of laser scanner systems 
some systems deliver a single reflectance value derived from 
the intensity or the power of the backscattered laser light. The 
intensity is determined by the signal maximum and the power 
by signal integration of the measured laser light and gives 
radiometric information about the surveyed area. This intensity 
(power) value can be used for separating segments of artificial 
objects from vegetation (Hug & Wehr, 1997; Maas, 2001) or 
perfectly textured 3-d scene models (Sequeira er al, 1999). 
Vosselman (2002) suggested considering the reflectance 
strength of the laser beam response to estimate and improve the 
accuracy of reflectance edge positions. 
In this paper we describe a simulation system and investigations 
for analysis of recorded laser pulses. In Section 2 a short 
overview on the simulation setup we used is described. The data 
generation is presented in Section 3. In Section 4 the processing 
chain for the data analysis is explained. 
2. OVERVIEW OF THE SIMULATION SETUP 
We focus on the aforementioned problem of measurement 
situations on building boundaries. A simulation setup that 
considers simulated signals of synthetic objects was build up for 
exploring the capabilities of recognizing urban objects using a 
laser system. The simulation setup is split in data generation 
which is described in Section 3 and data analysis in Section 4. 
A schematic illustration of the simulation setup is depicted in 
Figure 2 
The data generation can be carried out by a simulation or a 
measurement of the temporal waveform. By the use of a 3-d 
the building model (Figure 2-1) and 
the extrinsic orientation parameter for sensor position and 
orientation (Figure 2-2), the model is sampled to get a high- 
resolution range and reflectance image (Figure 2-3). The 
resolution has to be higher than the scanning grid we want to 
scene representation for 
simulate for further processing. Considering the temporal and 
spatial laser pulse description is relevant for modeling the laser 
pulse (Figure 2-4). To simulate the scanning of the laser 
system, the values of grid spacing and the divergence of the 
Internai 
laser b: 
image | 
resoluti 
(Figure 
data is : 
the rec: 
intensit 
the sigr 
what w 
To imp 
intensit 
with tl 
intensit 
measur 
each de 
with pu 
the ilh 
determi 
algorith 
descrip 
the san 
power 
edges e 
pixel 4 
receive 
to be cc 
For sin 
pulses 2 
3.1 Sc 
3.1.1 
For a 3 
geomet 
the for 
reflecta 
homoge 
based o 
  
Figure 
3.1.2 
The obj 
ample 
l'ecause 
spatial 
and or 
resoluti