lits
ele
24°
4°
cal
lon
um
jen
the
ind
vas
ng
ich
ere
ues
lits
lial
ere
h),
ital
ind
ind
lir-
of
to
ien
"he
1ge
jon
in
Table 2 The radial lens distortion for each side vs. collimator shift angle.
(by EFL)
shift angle radial lens distortion(mm)
H V SIDEI SIDEII SIDE III SIDE IV
+a° + 4° 0009 0.005 -0.002 -0.005
£12’ +8° -0.005 -0.004 -0.013 -0.011
zig? +12 -0.049 -0.025 -0.066 -0.018
+ 24° +16° -0.127 -0.053 -0.144 -0.043
Table 3 The coefficients of radial lens distortion in each side obtained by
collimator test. ( xE-3)
coeff. SIDE I SIDE II SIDE II SIDE IV
0.003546 0.003174 0.001178 0.004410
-8.552213E-5 -1.996202E-4
K,
K,
K, 277313757
K,
-4.344807E-5 -4.285227E-4
2.024559E-6 -1.133125E-7 7.853835E-6
-4.608873E-10 -8.256863E-9
4.990564E-10 -4.291548E-8
TIT T7173
/
^A
—120
radial distortion (um)
1
|
l
/
//
dd
/
A
H EFL
—160
l I 1 1
0.0 3.8 7.7 11.7 16.0
radial distance (mm)
Fig.3 Radial lens distrotionby EFL and CFL
Analytical Plumb Line Method
Grid images are obtained by wide angle lens with
35mm focal length through double exposures. 10
plumb lines are imaged in first exposure and 7
plumb lines in second exposure. We obtained radial
and tangential lens distortion coefficients of Nikon
F801 non-metric camera.
SIMULATION TEST
Using the coefficients obtained by the camera
calibration, we performed the simulation test to
examine the accuracy of results through the error
correction.
Table 4 Lens distortion coefficient of Nikon 801
non-metric camera.( X E-6)
Dis; radial distortion coeff.
tance
(m) K, K, K,
0.5] 43.82876730|-0.03626487 |-0.00000921
1.0) 53.17592457]-0.07958205|-0.00005170
1.8| 42.51753570]/-0.03721015]-0.00001845
Dis: tangential distortion coeff.
tance
(m) P P. P,
0.5| -30.87500896/-17.43285964|-0.74695714
1.0| -20.93058837| 7.74351767|-0.01338893
1.8] 9.84737198| 25.69442983 |-0.28874270
The 35 targets arrayed on the plane wall were
photographed with 9 photos obtained at 0.5m,
1.0m, 2.0m with the convergent angle 30° and
normal case. The targets were rearranged for each
object distance in order to be covered uniformly
on the full film area.
The standard deviations of results of two casese
were compared between systematic error to be
corrected and not to be through plumb line
method and collimator method (Fig.5). It is noted
that compared with uncorrected case, both methods
have 2096-3096 decrease of standard deviations.