ISPRS, Vol.34, Part 2W2, “Dynamic and Multi-Dimensional GIS”, Bangkok, May 23-25, 2001 
197 
Table 5-3 computation result 
Interior 
orientation 
parameters 
xO(pixel) 
yO(pixel) 
F(pixel) 
-274.50 
-236.98 
2604. 76 
Exterior 
orientation 
parameters 
U) 
K 
-0.6910 
0.5642 
0.9172 
7üM 
Xs(mm) 
Ys(mm) 
Zs(mm) 
57910.90 
-47492.94 
54088.57 
Cij 
x(pixel) 
y(pixel) 
X m 
ym 
~Z nr ~ 
X 
Y 
Z 
1 
1.0007 
-217.71 
196.13 
-7673.37 
6912.71 
-91805.01 
-2.53x10° 
2.08x10° 
-2.36x10° 
2 
1.0001 
-215.74 
195.10 
-7599.46 
6872.26 
-91751.15 
100 
2.41x10" a 
-2.74x10' A 
3 
0.9994 
-216.07 
197.39 
-7605.50 
6947.91 
-91686.03 
100 
-2.83x10 s 
100 
4 
1 
-218.04 
198.42 
-7679.40 
6988.36 
-91739.89 
1.75x10 s 
-1.44x10 s 
100 
5 
1.0013 
-215.69 
197.48 
-7606.28 
6964.09 
-91858.48 
-1.21x10‘ fa 
100 
-1.13x1 O’ 6 
6 
1.0006 
-216.01 
199.76 
-7612.32 
7039.75 
-91793.36 
2.64x10' s 
99.99 
100 
7 
0.9999 
-214.04 
198.73 
-7538.41 
6999.29 
-91739.50 
100 
99.99 
100 
figure 5-1 
real test 
We used a piece of real image, The size of the image is 640x480 
pixels, as illustrated in figure5-2. 
figure 5-2 Original image used for test 
Firstly, edges of objects (e.g. box) in the image are detected 
semi-automatically, we select corners, which are really in 
intersection point of two or more straight lines, as distinct points. 
The experimental results are shown in Table5-4.Three rendered 
views of the reconstructed box is in figure5-3. 
Table5-4 calibration results of proposed method 
Xs 
(cm) 
Ys 
(cm) 
Zs 
(cm) 
<P 
U) 
K 
xO 
(pixel) 
zO 
(pixel) 
(pixel) 
-53.28 
-62.01 
21.75 
1.3635 
0.8825 
-1.4452 
-77.41 
6.30 
1628.24