Abstract: The aim of this paper is to introduce some examples of applications of an active
triangulation technique. This technique is one of the most used methods of optieal3-D
object measuring and reconstruction and has wide utilization in industry. There are
mentioned some significant attrihutes of this technique, description of working principle,
possible variants and so on. Copyright © 20051FAC
2. TRIANGULA nON METHOD According to the light source we distinguish these three
variants:
The techniques hased on triangulation are most widely
used. Wc distinguish the follo\\~ng variants, which look 1-D triangulation light point
very dillcrenl. but use same principle: 2-D triangulation light stripe
3-D triangulation - light volume
258
I
•
'I
I
Fig . 2 . Calibration object.
I for matrix members (1/, t2, ... • t9) determination is
needed to make calibration by known object (known
spatial coordinates of points on object. Fig. 2). Then
data filc with sct of points in format 11 , 1', X , y, z is
necessary to provide. With these data the
Fig. 1. Principle of the triangulation method
transformation matrix is establi shed by minimisation
(I-D technique).
of differences hetween set and computed values.
2.2 Formulas.
2.3 Accura<:l'.
,I.-fathematical descnjllioll of l-TJ triallgulatioll. If the
If the investigated ohject is situated further to the
relleeted ray of the light is projected to the ,/' pi xci
call1era and I ight source. the projection a is bigger
from the total number pxl of pixels. than the size of
too. This means that reflected light ray is projected to
the corresponding projection a [mm] can be
the pixel more right (according to fig. 1). In this
calculated by formula (1) (Kalova and Lisztwan.
way, the di stance of the object can he obtained just
2005). where c is the chip size [mm].
from the position of the light point in the image from
c·n thc camcra . Thc mcasuri ng accuracy and also thc
a= -- [mm]. (1) usahility of this method depend on several
pxl parameters. The beller resolution (smaller difference
of measured distances of two neighbouring pixels -
The projection a is used to angle fJ computation lesser discretisation error) can be obtained by
if .. focal length I mm J). increasing si ze of base b. camera resolution pxl
fi=a(.'lg(c/~-aJ+90" n. (2) (number of photosensitive elements) and objective
size c . Fine accuracy is achieved in shorter measuring
b. sin a . sin f1 If the light volume is used. the whole object can be
l = [mm] (3) marked en block so no scanning is required. This is a
sinCUW" - Ca + fi)) big advantage as against 1-D and 2-D techniques,
when the object ha s to be measured step by step . By
using variants with light point or stripe just onc point
Calibratioll. These formulas are very sensitive to
or profile is ohtained during one measuring (one
accuracy of inserted data. Small shortcoming on
imagc). Gauging of wholc objcct rcquircs scvcral
input side brings big inexactitude on output.
separated measuring with changed position of object
Mathematical description for 2-D and 3-D
or beam of light. On the other side. these variants arc
triangulation techniques are more complicated too .
usuall y faster (lincar or specifically adaptcd camcra -
That ' s the reason why a transformation formula
see '2.7) and structurally simple (cheaper). There are
obtained by calibration is usually used. The
advantageously uscd "ith somc complemcntary
transform matrix (dimension 3x3) is used . Eaeh
linear or rotary movement (f. e . on production line .
constituent mc III ber rcprcscnts mutual relation
where thc measured objcct moves in it self).
between image (coordinates u, 1') and object space
(coordinates x. y. z). This relation is characterizes by
following equation (4). 2.5 Impurlanl drawbacks.
259
eeo
/\3 /
L-/ _--'V'---_______I
Fig. 3. The disadvantage or method.
260
Fig. 6. Characterful detects on welding seam:
pinholes, bubbles, wrong width.
261
From the image are restored the surface cylindrical spatial
coordinates r. z of the seam. Coordinate rp is obtained from
angular position of sensor units. Spatial coordinates are
evaluated by defect-detecting algorithm. At the cnd of the
working cycle all required details on the welding seam are
available and a good /bad signal is transmitted to the main
control unit. Wheels with significant defect are weeded out
from the production line.
The camera-projector system has been designed as a single 3.4 r -elll'cle detection and high profile acquisition.
unit where the camera and the projector are fixed on carrier
plate . Their optical axes intersect at the focusing point. This The technique with laser beam (I-D technique) was used in
design makes focusing and traffic industry for vehicle detection and high profile
maintenance easier and it Illcds other tcclmical acquisition. The syslcm can measure vehicle length and it~
requirements of production line too. Two sensor units arc velocity too. The vehicle can he classified according to the
used together (Fig. 9). high profile.
The device. formed from the linear camera and four lasers
fixed on a platform (Fig, 10). is placed above the roadway
and monitors the area helm.. The measurement rate is given
by the sampling frequency of the camera (around 0,5
ms/sample). The high profile of the vehicle (Fig. 11 c; Fig.
11 a and 11 b show corresponding photo and camera data) is
acquired in this way. Multiple lasers are used to provide
sufticient amount of data for measurement on the whole
traflic lane width .
Fig. 8. a) Droplet image with marked high profiles. Fig. 10. Photo of sensor unit.
h) High map function h (xy).
262
~\
4 . CONCl.USION
ACKNOWLEDGEMENTS
REFERENCES
263