wheels. Together these sensors generate the absolute 
positions of the GPS-Van and its orientation (attitude) at 
any time. 
For relative positioning the stereo-vision system was 
installed. It consists of two fully digital CCD cameras 
(Cohu 4110) with a resolution of 732 x 484 pixels. They 
are mounted on a rack on top of the vehicle. We assume 
that they are rigidly attached to the van and do not change 
their attitudes during operations. The two cameras directly 
interface to a real-time imaging system (Trapix Plus from 
Recognition Concepts Inc. (RCI)), where the images are 
temporarily stored in a frame buffer. They can also be 
processed on-line using a digital signal processor, or they 
can be sent to the Data Store real-time disk, which has a 
data transfer rate of 4 MBytes per second and holds 2 
GBytes of digital data. It is interfaced to an Exabyte digital 
tape drive through a SCSI connector. 
Finally, a touchscreen is used to control operations of 
the data-collection procedure and to key in a number of 
pre-defined attributes as the GPS-Van passes by an object 
of interest. A color-video camera is applied for photo- 
logging of the road environment; the video scenes are also 
related to the GPS-positions. All sensors of the mobile 
mapping system are controlled by a PC. The vision-system 
configuration is shown in figure 2. 
Stereo-Vision System (RCI TrapixPlus) 
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
PCT sere [L1 ave DJA 
nterface LÀ 
à XDPI RAM converte 
= VISInd —7 Hard 
= Adapte À disk 
4 Pixel KRTP Bd Hard 
© Processor a 9 d 
VISIn = 9 
S KVPE Adapt > d TU. 
= | ERIP ane 
‘sb à [Digital Poli | l'a Brel 
b } Interface LL S MUN 
KDPI 
A1]. 
Interf. 
  
  
  
Figure 2: Hardware components of the GPS-Van. 
3. INTERIOR AND RELATIVE ORIENTATIONS 
By calibrating the vision system we determine 
parameters that define the camera geometry and the relative 
locations and attitudes of the camera-pair. The calibration 
consists of two components: the interior orientation, and the 
relative orientation. It should be repeated in regular 
intervals to ensure that the camera-setup did not change. 
The calibration is performed by analytical methods, which 
involve capturing images of known control points (test- 
field), measuring their image-coordinates, and computing a 
photogrammetric triangulation to obtain the specific 
parameters. Both orientations were combined for this 
special application and can be solved simultaneously. Once 
the calibration is available, any object in the field of view 
of both cameras can be positioned in three dimensions in a 
local coordinate system. The transformation of these points 
to global coordinates is discussed in chapter 4. 
The Interior Orientation describes the geometry of a 
camera and consists of the focal length (c), the principal 
point (xp, yp), and lens distortions. For each of the cameras 
a separate interior orientation must be determined. The 
Relative Orientation defines the tilts of the two cameras in 
a local coordinate system, which has its origin in the left 
perspective center, and its Z-axis is perpendicular to the left 
   
   
   
    
   
   
  
   
   
  
   
   
   
   
  
   
  
  
   
   
   
   
  
  
   
   
  
  
  
   
  
   
  
   
   
   
   
   
   
   
   
   
   
  
  
  
   
  
  
   
  
   
  
   
  
     
image plane. The relative orientation is scale-independent; 
itis defined by five parameters. 
The combined solution of interior and relative 
orientations was developed to ensure that the relative tilt 
angles of the two cameras as well as the camera geometries 
are kept constant for all stereo-pairs of the test-field. In 
general, it is important to acquire a number of stereo-pairs 
at different, oblique angles and distances from the test-field 
to ensure an homogeneous positioning accuracy of the 
stereo-vision system. 
External measurements are added to enhance the 
stability of the least squares solution. We used theodolites 
to identify the perspective centers of the cameras as the 
entrance pupils of the lenses. The distance between the 
perspective centers determines the base of the stereo-vision 
system. It defines the scale of the local system in which 3- 
dimensional points are positioned and must be very 
accurate. It is used as a constraint for bundle adjustment. 
Figure 3 shows a typical calibration set up of two van 
positions and the theodolite stations in front of them. In the 
following the analytical formulation of the combined 
adjustment is presented, and the physical meaning of each 
of the parameters is explained. 
building 
testfield of 
control 
points 
AL rera a uide 
  
^ 
theodolite stations 
  
Figure 3: A rigorous calibration of the stereo-vision 
system is achieved by a combined bundle adjustment with 
additional camera parameters and geodetic constraints. 
The analytical calibration of the vision system is done 
by the bundle method based on collinearity equations (1) 
(Brown, 1976). To determine the interior orientation 
parameters simultaneously, they are also treated as 
unknowns. This means that we compute the coordinates of 
the principle point (Xp, yp) and the focal length (c), in 
addition to the orientation parameters of each camera. We 
also solve for two parameters to model radial distortions, 
two parameters to model decentering distortions, and for 
two affine deformations (2). 
Collinearity equations: X=-C = + Ax (1) 
N 
y 
-—C—-A 
y °D y 
XY. Dinh image coordinate measurements, 
ERE MG AA e focal length, 
Ny, Ny, D....... numerators and denominators of the 
collinearity equations.