Chikatsu, Hirofumi 
  
+ Tracking is controlled. 
^ Monitor part 
+ Stereo image with times are displayed on VTR 
monitor. 
* Time data, pan and tilt rotation angles and 
tracking image are displayed on PC monitor 
Then, the synchronized stereo image sequences, 
times, pan and tilt rotation angles can be 
obtained in realtime while tracking a moving 
object. Figure 3 shows the VTR monitor image. 
The number of right lower corners (03' 44.67") 
shows the superimposed time. 
3 CAMERA CALIBRATION 
Figure 4 shows coordinate system for the HVT 
system. The unknown parameters for the exterior 
orientation parameters {u, J) ép, (rotation 
  
parameters), X», ¥), Z (camera positions)! and the X, Y, Z : Body coord. system 
interior orientation parameters {f (focal length), x, Z X^ Y Z': Head coord. system 
Jo (principal points), a;, a», (scale factor), p; (lens X, y,Z : Camera coord. system 
distortion); for the CCD1 and CCD2 camera can be 
calculated by the same procedure in basically for the Figure 4. Coordinate system of HVT 
video theodolite with one target (Huang and Harley, 
1989). 
One reflector as a target was used to measure object distance above datum. The most remarkable point of this HVT 
system, however, automated camera calibration can be achieved without any reflector since object distance can be 
measured by the laser range finder. The brief calibration procedures are as follows: 
When the CCD1 and CCD2 camera are pointing to Py point under the condition that the distance is D from the 
center of the body part of the HVT to Py, horizontal angle H and vertical angle V;, the image point to Pg is taken as 
po on the image respectively. Next, when the HVT is rotated V, angle, the image point to Pj is taken as p;, and the 
object position to image point p; becomes Pj. Similarly, when the HVT is rotated H; angle, the image point to Py is 
taken as p», the object to p; becomes P. Repeating this operation, control points are taken on the image and are 
produced in the space respectively. 9 control points were produced in this paper, and the ground coordinates for 
these control points are calculated with following equation, 
X; cos(H;) sin(H;)sin(V;) sin(H jJcos(V; ) cos( H, ) 0 - sin(H, ) 0 
Y; |= 0 cos(V;) —sin(Vi ) sin( H, )sin(V, ) cos(Vy) cos(H, )sin(V, ) || 0 ( 1) 
Zi -sin(H;) cos( H;)sin( Vi) cos( H;)cos(V;) || sin( Hy )cos(Vy) —sin(V,) cos(Hy )cos(Vy) || D 
Where, these ground coordinates are opposite the body coordinate system in Figure 4, and H;, V; are measured 
clockwise. 
Furthermore, when the stereo image are taken by rotating the body V; and head H;, each camera position has to be 
corrected to respond to the rotation of the HVT by the following equation since the stereo image are taken at a 
different exposure station due to the discord between the center of the body and the lens. 
1 Fz a 0 cos(Hj) 0 sin(Hj)|[Xo 
Yo" |=|0 cos(V;) —sin(V;) 0 1 à Y, 
(2) 
Zu. 0 sin(V;) cos(V;) —sin(H ;) 0 cos(H; ) Zp 
* * * "e. 
where, X; , Y; and Z, are corrected camera position. 
  
  
132 International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B5. Amsterdam 2000. 
  
It has be 
sum of c 
Howeve 
rotation 
head Hj! 
i=. 
where, 
€gy € 
€3 € 
€3] € 
and 
405 ai 
03] Si 
4 AQI 
In ord 
followi 
using 
test fiel 
The tai 
values 
Black z 
in calib 
check y 
+Test1 
Camerz 
0mm p 
HVT r 
taken : 
Similar 
mm, + 
the H\ 
taken. 
Test - 
On th 
perforr 
image 
stereo 
the H\ 
A3 we 
stereo 
mm on