re corrected using the
r— Xo) ed,
(18)
coordinates of a pixel
it;
es of the same pixel in
dinates of the principal
| to the principal point;
dial distortion (higher
entring distortion are
atures
ge have been corrected,
plane can be expressed
with the least square
n.
L RESULT
al have been conducted
ind the effectiveness of
eme of camera. The
ch were extracted from
, were used to describe
hod. The real date was
the point-based method
re. For the former, the
x. Figure 3 illustrated
| listed the simulated
nulated camera.
3(d)
trol Target.
Target.
( b ). Edges Detected By Sobel Operator.
( c). Edges Extracted By Thinning and Following.
( d). Straight Lines (black frame) Extracted with the
Least Square Fitting Method.
Table 1. Simulated Valuse of Orientation for Camera
Interior Xo Yo f
Orientation
Parameters| —3.692 2.972 699.42
Xs Ys Zs
Exterior | 54410 -628.04 1555.90
Orientation
Parameters
® K
53.559803 57.089316 349.864286
Table 2 listed the simulated parameters of control lines
and their observations on the image plane.
Table 2. Simulated Values of Control Lines
p 9 Xc Yet a B 1
LO 10.10 0.164635 100.00 100.00 0.00 0.000 0.000 1.000
188.19 4.351133 100.00 0.00 0.00 0.000 1.000 0.000
12 154.08 5.733119 0.00 100.00 0.00 1.000 0.000 0.000
13 39.68 4473506 100.00 0.00 600.00 0.000 1.000 0.000
LA 241.34 3.541581 100.00 1100.00 0.00 0.000 0.000 1.000
[5 172.91 2.159794 0.00 1100.00 600.00 1.000 0.000 0.000
L6 42.58 5.585684 0.00 100.00 600.00 1.000 0.000 0.000
L7 238.44 6.233983 2100.00 100.00 0.00 0.000 0.000 1.000
235.29 1.558492 2100.00 0.00 600.00 0.000 1.000 0.000
The software was programmed with C++ language
and executed on PC in the environment of MicroSoft
WINDOWS 95. The whole process consisted of : ( 1 )
optical distortion parameters; ( 2 ) principal point
coordinates; ( 3 ) affine scaling parameters; Assume the
system errors of image are stable and corrected and the
principal point coordinates are unknowns, and (4)
exterior orientation parameters for camera. The control
lines can be automatically extracted by the program at
sub-pixel precision.
Table 3 lists the initial values and the calculated
result.
Table 3. Result of Single Photo Resection
Initial Value Calculated Value
X, -3.692 ---
Yo 2.972 ---
f 699.42 ---
Xs -930.100 -934.127
Ys -620.041 -628.043
Zs 1550.000 1555.880
Q 50.602520 53.560490
o 60.141391 57.088574
K 359.884333 349.864290
In order to compare the new method with other
point-based ones, we used a piece of real image, which
was taken with film-based 120 camera and scanned at
resolution of 600 DPI (the original image shown in
figure 4). The control points were measured manually
with cursor on computer screen and listed in table 4.
Two point-based methods DLT (Direct Linear
Transformation ) and SR ( Space Resection ) were used
and the results are listed in table 5.
Figure 4. Original image used for test
Table 4. Coordinates of Control Points
Point No. x [pixel] y [pixel] X[m] Y[m] Z[m]
1 196 915 6132.873 . 855.600 115.477
à 241 988 6125.498 850.850 105.256
3 244 1257 6125.608 850.856 70.128
4 721 441 9/94.480 | 875.060 110.521
5 1854 513 5154.531 189.101 110.505
6 2353 518 5754.602 746.094 108.206
1 497 1631 5754.504 891.241 . 28.960
8 877 1623 5754.462 | 864.894 28.960
9 1991 1598 5754.471 778.866 28.950
10 735 1744 5154.521 874.942 20.112
11 2879 1353 5823.379 649.903 49.288
12 3178 1265 5783.367 644.957. 56.144
13 3553 1335 5764.857 617.183 49.318
14 2864 1698 5823.420 652.165 16.158
15 3599 945 5821.447 560.807 91.456
16 3892 882 5764.506 575.348 | 91.460
17 2485 1711 5605.802 | 824.259 | 26.047
18 2516 1709 5606.329 822.147 26.044
Table 5. Result of Single Photo Resection
DLT Method _ Point-Based SR
X 83.296 83.396
vas -57.371 -57.371
f 239.493 239.494
Xs 5366.991 5366.991
Ys 966.396 966.396
Zs 36.896 36.896
Q 115.601898 115.601904
® 91.394714 91.394786
K 359.140747 359.260951
Using these control points we could get several
