0.8 r
|
|
|
0.6 Fi
I
ln
= "
5 il
2 i — - — Patch radius 10pixels
a i Patch radius 8pixels
704}; ID Patch radius 6pixels
= - Patch radius 4pixels
= Patch radius 2pixels
©
&
EaSMN ES
a Ls ;
Eigen values
Figure 11: Probability of eigen value with patch dimension.
Table 2: Deviation of interdot angles of the laser projector.
No. Min. | Mean | Max. | r.m.s. | Std dev
gon gon gon gons gon
361* | 0.00 | 0.008 | 0.047 | 0.016 0.013
361** | 0.00 | 0.010 | 0.047 | 6.013 0.008
* — Vertical angle & ** — Horizontal angle
which uses a combination of a camera and a projector, these
parameters for the camera are calculated using a standard
test field or by fixing a CCD camera rigidly to the theodo-
lite of a geodimeter [Huang & Harley, 1990]. The use of a
projector as a second sensor creates a compatibility problem
unless the projector is not treated like a camera. The pro-
jector makes it easy to solve the correspondence problem but
three dimensional calculation of object co-ordinates becomes
a little difficult or rather different from standard photogram-
metric procedures due to the unavailability of the orientation
parameters of the projector in a format of the real camera.
In establishing the reliability of the dot matrix projector, vir-
tual three dimensional testfields were generated using two sets
of observations of the geodimeter for two different distances
from the fixed autoreflecting spherical target. Corresponding
to each of these control targets an image co-ordinate was
derived through the known regular pattern of the diffraction
grating based dot matrix. The known positions of control
points and their corresponding image co-ordinates were used
for resection in the combined adjustment program (CAP) to
calculate interior orientation parameters of the projector. The
bundle adjustment results for the camera model of the laser
dot projector were comparable with those of a real CCD cam-
era (Table 3) which were derived using a standard three di-
mensional testfield. Due to the narrow field of view of the
projector, the value of X and Y co-ordinates of the principal
point were found to be highly correlated with phi (rotation
about y-axis) and omega (rotation about x-axis) respectively.
The lens distortion model of the laser projector showed rel-
atively higher values of symmetric radial and tangential dis-
tortion but smaller values for decentric distortion compared
Table 3: Standard deviations of interior orientation parame-
ters of the projector and the CCD camera.
Parameters CCD camera | Laser projector
Pixel Pixel
Focal length 0.026 0.014
(Std dev)
PP co-ordinate 0.026 0.063
(Std dev)
Ki 0.475D-07 0.596D-04
K» 0.543D-13 0.851D-07
Ks 0.932D-19 0.137D-09
Bi 0.195D-01 0.860D-03
B5 0.123D-04 0.633D-03
to those of the CCD camera. The residuals after bundle ad-
justment for interior orientation parameters measurement of
the CCD and projector are shown in Figure 12.
> b 9 $
One pixel: ————————— 5
} f
e od el > | o
$ e e d
> LP o "d
d o a ©
v No ef o
> 9?
© o 9 ©
RK d 2 e >)
A axXIS
ds > One pixel: —
do OO e s a
OO o Qo o Qo oo o
Ping aies ee
Do o is Se
99 009
SO og o ooo
OO oa 20060
ee oos 9 €9 a.
MI
Ra
ooo a
oo
eoo
eoo 17906
© oo
SC eed 99 a
8e eo
ee ee
Sons ong o
en Lem eog s e
Oo o mong Te
eoi SR 0
oe. OO en o o0
200 © © 2 Qe
o
e 00 Q4
o Oeo o
oo
eo ©
Siro emo,
999511290041 °
Oed were. ec
© ooqQ]
— Y-axis
©
Q
©
G
0
G
Figure 12: Residuals after bundle adjustment for camera and
projector.
4 OBJECT SPACE MEASUREMENT
It is important to find the relative orientation of the projec-
tor with respect to the camera to obtain spatial intersection
among the projected laser dots and their respective image
co-ordinates. Both camera and projector are placed on the
telescopes of geodimeters, so the relative orientation of one
with respect to the other is measured by mutual pointing of
the telescopes [Allan, 1993]. The distance between the two
levelled geodimeters and the observations of their vertical and
horizontal angles after mutual pointing are used for scaling
544
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B5. Vienna 1996
the |
recti
obse
all tl
the |
poin
rays
was
4.5 1
Simi
get |
prov
turel
jectc
accu
mea:
Thet
cali
in tl
drive
obje
over
able
used
mea:
