T
' Ito,H. Aoki
photogrammetry
surement". It is the
ite 3D measurement
ersified in its form,
gn, or sandy with
hich characteristics
hen an object to be
ry.
ution CCD cameras
PI-1000. And its
o-fine correlation".
res, the perspective
'00 can be output in
simulated surfaces
n the flat surface of
of a bullet train of
‘and their inputting
‚its principal point
mera lens, we have
w-cost lens whose
inge even with the
igh resolution CCD
1e calibration only
NO son USG ter the
istment software",
ve also developed,
letect automatically
fied designed for
irm and verify the
rement devices we
f experiment. First,
5 measurement we
indy and lustrous
1 the actual surface
Second, for curve
d on the simulated
d our method on the
€ will first present
stem, and then its
by the examples of
stems.
na 1996
2. OVERVIEW OF THE SYSTEM
The surface measurement system we have
developed is an analytical stereoplotting system
for digital image. As shown in Fig.1,2 the system
consists of an image-acquiring part and
image-analyzing part.
The image data that are obtained by taking a pair
of left and right stereo images are stored in the
separate optical or hard disks for respective side
and then transferred into a stereo-image
workstation (PI-1000) and analyzed therein.
2.1 Image-Acquisition
The image-acquiring part is composed of two
high-resolution CCD cameras, a pattern texture
projector and base-bar. The CCD cameras and
projector are placed on a base-bar and can be
easily moved about in a unit to acquire images on
the surface of a vehicle's body.
Our high resolution CCD camera has 4096 X
4096 pixels and is made such a way that even if
the focal length is changed, its principal point
does not fluctuate.
And this time, for camera lens we have developed
a special and low-cost lens whose focal length is
29mm, maximum aperture is 4, and the view angle
is 68^ with its peripheral distortion contained
within the range of 0.05%. With the former metric
camera, we were obliged to carry out the
camera-calibration each time the focal length had
to be changed or the lens had to be replaced. Our
new high resolution CCD camera, however,
requires the calibration only once, even though the
focal length changes with different objects .
The Fig.3 shows the fluctuation of the principal
point with the change of the focal point. Here the
fluctuation range is contained in less than one
pixel (CCD pixel size : 7u m square).
X-Accuracy
(ym)
20
of
Q1 Md ——— 3
i 0/502. 9 3504,05 05/,07.08/159 1
Focal point(mm)
E
=
—
-10
-20
Y-Accuracy
Cum) =
20 =
r-|
4 =
0 j i n zs | ei
T t t
| |
T 0.1 3.2. .074 0.4 1.5 0.6 0.7 0.8 0.9 1
-20 Focal point (mm)
Fig.3 Fluctuation of principal point
Fig.1 Image-acquiring part
Fig.2 Image-analyzing part
2.2 Image-Analysis
For image-analysıs we used the stereo-image
workstation PI-1000, which we had developed as a
digital plotter.
PI-1000 is an apparatus which conducts 3D
measurement and plotting, employing the image
information obtained through a high-resolution
CCD camera (PC-1000), the image scanner
(PS-1000) which we had developed and other
CCD cameras. This apparatus, in fact, has a wide
range of applicability for 3D measurement, such
as civil engineering and industrial measurements
etc.
The 3D measurement is conducted as follows.
First, when there is no particular texture or
design on the object to be pictured, we project on
the object a pattern design by pattern-projector
and take its ‘digital picture by à pair of
high-resolution CCD camera. And we transfer its
data now stored in optical disk or hard disk into
the, memory of . PI-1000. PI-1000 then
automatically rectifies the stereo image by
rearranging the image data of right and left side
on the epipolar line with the distance between the
two cameras on base-bar as its base-length.
Next, the rectified image of the right and left
side is displayed on a 3D image display, on which
the floating marks and measurement results are
superimposed.
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B5. Vienna 1996
