vhich 
ding 
data. 
nable 
ilding 
> data 
range 
usion 
ting a 
ssured 
h can 
close- 
imeras 
images 
mation 
; about 
tarting 
points 
results 
)), and 
jentary 
ich that 
uilding 
ind the 
1 future, 
ped to 
points 
nearly- 
was l0 
ystems. 
systems 
Ve then 
ipplying 
ata sels 
sting of 
| section 
ectively. 
plication 
ummary 
International Archives of the Photogrammetry, Remote Sensing 
Reference 
Data 
Acquisition 
Sensors ll Data 
Terrestrial i Surface 
Laser Scanner => i Points 
Digital Camera => | Images 
: f 
i É 
Total Station —> i 
: Points 
Figure 1. Framework for generating precision 
2. THE PROPOSED FRAMEWORK 
We propose a framework to generate precision building models 
from terrestrial sensory data. As shown in Fig. 1," this 
framework includes two main stages, data acquisition and data 
processing. Data acquisition involves collecting three types of 
data about a target (a building). These data are surface points, 
images, and reference points acquired by a laser scanner, a 
digital camera, and a total station, respectively. Data processing 
involves deriving a precision building model from these sensory 
data. The model retains geometric information (shapes and 
positions) in absolute coordinates with radiometric information 
(texture). 
3. DATA ACQUISITION 
3.1 Sensor Overview 
3.1.1 Laser Scanner: Using a laser scanner, we can acquire 
numerous points densely sampled from the surfaces of a 
building. The laser scanner transmits laser pulses to the target 
and receives the return pulses reflected at a surface of the target. 
It then computes the travelling time of a laser pulse and 
converts it into the range between the scanner and the target. 
The range is combined with the direction of the pulse recorded 
by the scanner itself to generate three dimensional coordinates 
of the point on the target at which the pulse is reflected. The 
coordinates are expressed in a coordinate system fixed to the 
laser scanner. The coordinate system depends on the orientation 
and position of the scanner and thus is not an absolute 
coordinate system fixed to the Earth. 
3.12 Digital Camera: Using a digital camera, we can 
acquire. color images of a building. This camera is not 
necessarily a metric camera retaining high resolution and 
accuracy. A general purpose digital camera must be sufficient 
since the images are used for obtaining only texture information. 
To derive precise geometric information from the image, we 
should first perform camera calibration processes. A promising 
algorithm for the calibration is presented by Habib et. al. (2002). 
985 
and Spatial Information Sciences, Vol XXXV, Part B7. Istanbul 2004 
Data 
Processing 
Model 
Geometric Models 
(shape & positions) 
Radiometric Models 
(texture) 
Georeferencing 
(absolute coordintes) 
pos J 
building models from terrestrial sensory data. 
3.1.3 Total Station: Using a total station, we can acquire the 
accurate coordinates of reference points in an absolute 
coordinate system. A prism rod is erected at each reference 
point to facilitate a convenient measurement of its location 
using the total station. The measured reference points should be 
also identifiable from the point clouds and images acquired by 
the laser scanner and digital camera so that they can be used for 
geo-referencing these data. 
3.2 Sensor Configuration 
Sensor configuration considers positioning sensors and locating 
reference points. A filed of view of a sensor is always limited 
and thus the data set acquired at a position can cover only 
partial area of a building. Hence, the sensors should be carefully 
positioned so that the entire facets of the building can be 
covered. Data sets obtained at adjacent positions should include 
sufficient overlaps for their registration. Reference points are 
located so that each data set can include four reference points 
that do not lie on a straight line. Two points among them are 
overlapped by an adjacent data set. Figure 2 shows an example 
of sensor configuration, where the diamonds indicate the 
locations of the laser scanner and digital camera and the 
triangles those of the reference points. 
  
  
  
  
  
  
  
+ » 
\ A A 
A A 
A Aa —j 
/ A ^A \ 
/ \ 
§ i^ i 
/ a Y 
/ po a — te, à 
/ — " a A 
+ " $ 
Figure 2. An example of sensor configuration