ibul 2004 
ded on a 
RU. It is 
ication of 
inmanned 
onducted, 
or getting 
asons; to 
ition and 
are fixed 
'cometric 
integrate 
oordinate 
ressed in 
> interior 
ength (f), 
camera 
. Camera 
g control 
iputed in 
8 
x digital 
Eq. (2) 
on. 
  
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part BS. Istanbul 2004 
  
Xe =X +X (Kr) (1) 
Yu = y + x (K,r) (2) 
where X'=x—-k; Y =-(y- yo) DP-x4 y 7 
(x, y): image coordinate 
2.2 Exterior Orientation 
Calibration of digital camera and laser scanner is conducted to 
estimate exterior orientation parameters, positions and attitude. 
This needs shift and orientation between each of sensors and 
[MU, which is computed from calibration. At first, rigorous 
geometric relationship between laser scanner and digital camera 
is established. This relationship is strongly fixed at all times. 
Then, geometric relationship between digital camera and IMU 
is established. That is, geometric relationship between laser 
scanner and IMU is theoretically established. Figure 3 shows 
the concept of calibration to decide exterior orientation 
parameter. 
For laser scanner calibration, the solar cell is used. The solar 
cell connect to an earphone. If the solar cell receive laser beam 
from laser scanner, the earphone beep. In this way, the center of 
laser beam is measured for calibration even though laser beam 
is invisible. 
Laser Scanner Digital camera 
  
      
  
R:Uo,9, KU 
Six, y,ZU 
= <> 
ir Ri: 'oi,pi,xi! 
N Rılt) Si: LXLyLZI 
X Sit) 
RN 
IMU Total Station 
(common referencing) 
where R: Rotation Matrix 
S: Shift Vector 
Figure 3. Concept of calibration for exterior orientation 
Geo-referencing of laser scanner data and digital camera 
image is done by computing the rotation matrix, R, and shift 
vector, S with IMU. All the points scanned by the laser scanner, 
X, and digital camera, (x,, y,), in terms of local coordinate 
System is converted to world coordinate system as given by 
Eq.(3) and Eq.(4). 
Xd] = (Ri*R) xc (Si+S) (3) 
NNNM ue = f(Ri, Si, f. Xu» Yu) (4) 
Where — f(): collinearity condition equation 
3. CONSTRUCTION OF DSM 
3.1 Integration of Positioning Sensors 
The positioning and attitude of sensors are decided by 
integration of GPS, IMU, and images. One of the main purposes 
of this research is to integrate all sensors and to develop the 
high precision positioning system in all circumstances. 
Figure 4 shows the algorithm for deciding attitude, velocity, 
position using only IMU. 
  
  
  
  
  
  
  
  
  
  
  
  
Extra 
Delta Attitude & Delta 
Angle Yaw Velocity 
Initial 
Attitude 
  
+ y 
Quaternion Euler Matrix Coordinate 
Update » Generation P! Transformation 
  
  
  
  
  
  
  
  
  
  
  
  
Velocity Generation 
Coliolis Compensation 
Gravity Compensation 
  
  
  
  
  
IMU 
velocity 
  
  
  
Torque Rate 
Generation 
  
  
  
  
  
IMU Initial Position 
position Initial Wander 
  
Relative Rate 
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
  
Azimuth Generation 
v 
= e i / ei C 
Earth Rate eg 
Generation ander Azimuth — 
Generation 
  
Figure 4. Pure navigation block diagram 
IMU has a rising quality, but it is still affected by systematic 
errors. Through Kalman filter operation, an optimal estimate of 
the sensor position and attitude are determined from GPS and 
IMU. Meanwhile, relative orientation of CCD image is 
determined by bundle block adjustment. Bundle block 
adjustment is a non linear least squares optimization method 
using tie-points of inside block. GPS and IMU allow automatic 
setting of tie-points and they reduce the number of tie-points 
and searching time of tie-points by the limitation of searching 
area. GPS, 1Hz, and IMU, 200Hz, with Kalman filter which has 
high frequency is necessary for geo-referencing of range data, 
which frequency is 18Hz. But accuracy is low because of miss 
alignment, drift error, or effect of PDOP. On the other hand, 
CCD images are taken every 10 second and geo-referencing is 
computed by bundle block adjustment which has very high 
accuracy of mapping. But frequency of computation is very low. 
That is, because of different frequency and accuracy, these two 
different methods of geo-referencing, Kalman filter and Bundle 
block adjustment. 
Therefore, the combination of bundle block adjustment and 
Kalman filter is proposed in this research. The result of bundle 
block adjustment aids Kalman filter by initialization of position 
and attitude. In this research, the result of bundle block 
adjustment is treated as true positioning values. IMU and GPS