High-precision triangulation sensing of mirror surface

Sergei V. Mikhlyaev∗
Institute of Automation & Electrometry

ABSTRACT

The analysis of measurement errors of distance up to a mirror surface by laser triangulation is given. We consider the
two types of sensors: with output plane, orthogonal to the optical axis, and with tilted one (corresponding to the
Scheimpflug principle). The conditions of Scheimpflug principle applicability for mirror surface testing are determined. We
have shown, that at the certain ratio between parameters of optical system of the sensor and the probing light beam, the
sensor with orthogonal plane of photoregistration can provide smaller level of measurement errors, than sensor with tilted
plane, despite of the presence of defocusing.

Keywords: triangulation, laser, mirror surface, distance, measurement error, defocusing, gaussian beam.

1. INTRODUCTION

The optical triangulation is widely used in scientific researches and in different areas of an industry for measuring the
distances up to an analyzable surface. A common triangulation principle is to project a light spot onto the object and extract
the distance information from the reflected or scattered light. Mainly the two types of triangulation sensors are used: in the
first one (OP sensors) the output plane of a sensor is orthogonal to its optical axis, in the second one (TP sensors) the output
plane is tilted to the axis under a particular corner according to the Scheimpflug principle1. The tilting of an output plane
allows to eliminate a defocusing and to increase an effective range. Most efficiently these advantages appear at sensing of a
diffuse surface.
In this paper the conditions of Scheimpflug principle applicability for testing the mirror surface are investigated.
We have shown, that at the certain ratio between parameters of optical system of the sensor and the probing light beam,
the OP sensor can provide smaller level of measurement errors, than TP sensor, despite of the presence of defocusing.

2. PRINCIPLE OF MEASUREMENTS

The principle of triangulation measurements is explained in Fig.1. It shows the probing light beam with a central ray L
projected on a mirror surface P0 (z = 0) at an angle α. The reflective ray L1 coincides with the optical axes of the sensor.
The output (detection) plane of a sensor is perpendicular to an optical axis (P1), or is tilted under an angle β (Pβ) according
to the Scheimpflug condition. The distances d1 and d2 are connected by the optical conjugate condition: 1/ d1 + 1/ d2 =
1/F, where F is the focal length of the lens. Position x1 of a reflected beam L2 in a plane J1 depends on the coordinate z=h
of a tested point m of a surface P:
x1 = 2hM sin(α ) , (1)
where M=d1/d2 is the magnification of the optical system
According to the Scheimpflug condition and by using Fig.1 we can write the expressions for TP sensor:

tg ( 2α ) = Mtg ( β ) , x β = ∆ 2 / cos( β ) , (2)

where
∆ 2 = M 2 ∆ 1 /(1 − M∆ 1 / F ) , ∆ 1 = h cos( 2α ) / cos(α ) .

The measurement errors of a surface profile are determined by the measurement errors of reflected beam positions in output

∗
[email protected]; phone +7(3832) 34-34-53; fax +7(3832) 33-38-63; Institute of Automation & Electrometry, Pr. Koptyuga 1,
Novosibirsk, 630090, Russia

Optical Engineering for Sensing and Nanotechnology (ICOSN 2001), Koichi Iwata, Editor,
400 Proceedings of SPIE Vol. 4416 (2001) © 2001 SPIE · 0277-786X/01/$15.00

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 07/03/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx

 planes P1 and Pβ. From (1) and (2) follows:
∆h1 = ∆x1 2 M sin(α ) , (3)

∆hβ = ∆x ' β (1 − M∆1 / F ) 2 cos(α ) M cos(2α ) M 2 + tg 2 (2α ) , (4)

where
∆x ' β = ∆x β sin( β ) .
The measurement errors of the light spot position in turn depend on measurement technique, parameters of a signal and
photodetector.

3. ESTIMATION OF MEASUREMENT ERRORS

For estimation of measurement errors we used the following assumptions:

1. The intensity of a light beam is described by the Gaussian function

I ( x, y ) = (πσ 2 ) −1 exp(− ( x 2 + y 2 ) / σ 2 ) , (5)

where the width of a beam σ depends on a coordinate ξ along its axis:

[
σ (ξ ) = σ 0 1 + ( λξ πσ 02 ) 2 ] 1/ 2
. (6)

2. The N=SNR, determined by a dynamic range of a signal and a photoreception device, is limited:

N ≤ I max ∆I ,
where I max is the maximal magnitude of a signal at given measurement range.
3. The photodetector size is small.
4. The coordinates of a light spot are determined by measuring the front positions of appropriate electrical signal in the
points of maximal sensitivity.
On these assumptions we can write the expression for the computation of measurement error of light spot position:

∆x = I max NI x' ,
where I x' is the derivative with respect to x.
For Gaussian beam (4) at points of maximum derivatives xm = ± σ 2 we have
∆x = πI max cσ , 3
(7)
where c ≈ 1.2 N .
In order to obtain the required estimations, in (7) it is necessary to use the expressions for parameters Imax and σ for
appropriate measurements schemes. These expressions can be deduced from geometry of a probing Gaussian beam,
reflected from a surface z=h, as shown in Fig. 2. The beam waist transits through the point x=y=z=0. The sizes of a light
beam σ0, σh, and σ1 in planes I0, Ih, I1, orthogonal to the optical system axis, are given by expressions:

σ 0 = σ (0) , σ h = σ ( h cos(α )) , σ 1 = σ (2h cos(α )) .

The maximum values of signals for two considered measurement schemes can be written, according to (5), as

I β (h ) = (π ( M hσ h ) 2 ) −1 , (8)

Proc. SPIE Vol. 4416 401

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 07/03/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx

 I 1 ( h ) = (π ( Mσ 1 ) 2 ) −1 , (9)
where
M h = M /(1 − M∆ 1 / F ) .
From (8), (9) it is possible to receive:
1
I max = max{I 1 (0)} = 1 πσ 02 M 2 , (10)

β
I max [
= max{I β ( h )}= I β ( h0 ) = = 1 + (πσ 02 λ ) 2 ( M cos(2α ) F ) 2 πσ 02 M 2 , ] (11)
where
h0 = −(πσ 02 λ ) 2 M cos(2α ) cos(α ) F , (12)

From (3), (4), using (10), (11), we can write the expressions for ∆h1 , ∆h β :

∆h1 = c σ 0 (1 + ( λ πσ 02 ) 2 4h 2 cos 2 (α )) 3 2 2 sin(α ) , (13)

∆h β = c σ 0 (1 + ( λh πσ 02 cos(α )) 2 ) 3 2 (1 + (πσ 02 M cos(2α ) λ F ) 2 ) K , (14)

where
K = 2 sin(α )(1 − Mh cos(2α ) F cos(α )) .

Let's note, that (13) does not depend on magnification M of a system. The expression (14) corresponds to unlimited
measurement range. For a lens with the aperture D the measurement range is limited to a condition

h ≤ h D = D 4 sin(α ) , (15)

therefore the maximum value of signal should be selected from this effective range:

I β ( h0 ), for h0 ≤ hD
I~max
β
= I β ( − h D ), for h0 < − hD (α < π / 4) (16)

I β ( h D ), for h0 > hD (α > π / 4)

Using (16) it is possible to convert (14) to the form

∆h β = πc σ 03 M 2 I~max
β
(1 + ( λh πσ 02 cos(α )) 2 ) 3 2 K . (17)

4. SIMULATION RESULTS

In Fig. 3 the graphs of error ratio T ( h ) = ∆h1 / ∆h β for λ=0.63⋅10-3, D=50, F=70 mm, N=50 and for different values of
parameters are shown. As follows from Fig. 3, T (h ) can be greater or less than 1, depending on the values of parameters.
In particular, at σ0=0.2 mm, M=1.2 and α=20 the measurement errors for a OP sensor are less than the errors for a TP
sensor. The absolute values of errors are shown in Fig. 4. Let's note that in order to obtain the more precise estimation of
measurement errors it's necessary to take into account the distortion of light distribution in Gaussian beam due to the surface
inclination. However the accounting of this effect gives not very essential corrections to measurement results2.
Also it is necessary to note that the obtained results are related only to the case of a flat surface with fixed direction of its
normal n. If during the measuring the declination of a surface varies considerably, it can produce the essential measurement

402 Proc. SPIE Vol. 4416

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 07/03/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx

 errors in OP sensor. Nevertheless, this scheme can be used for testing the non-stationary surface and for correct statistical
evaluating the basic surface parameters3.

5. CONCLUSIONS

In triangulation sensors with tilted output plane (corresponding to the Scheimpflug conditions) there is no defocusing,
but they have nonlinear longitudinal (range) magnification decreasing the accuracy of measurements. In turn, as shown in
this paper, in sensors with an output plane orthogonal to the optical axis the measurement errors depends on a divergence of
a probing light beam. By using both of these factors we can see, that at sensing the surface with narrow scattering indicatrix
(mirror surface), the sensor with a photoregistration plane orthogonal to the optical axis can provide a smaller level of
measurement errors, than the sensor with a tilted output plane. These advantages take place at the certain ratio between the
parameters of the optical system of a sensor and the probing light beam divergence and remain valid under the small surface
tilts. Such conditions can be created by using the low-pass filter (diaphragm UU in Fig. 1) to restrict the angular deviations γ
of the reflected light beam. In this case more precise measurements can be fulfilled in local areas of a surface corresponding
to small values of γ. The sensors with output plane orthogonal to the optical axis can be used also for obtaining the
statistical evaluations of basic parameters of a nonstationary reflective surface3.

REFERENCES

1. R. Baribeau, M. Rioux, "Influence of speckle on laser range finders", Appl.Opt., 30, 20, pp.2873-2878, 1991.
2. V. V. Vertoprakhov, "Influence of the shape of an object and the orientation of its surface on the accuracy of laser
triangulation measurements", Optoelectronics, instrumentation and data processing (Avtometriya), 6, pp.61-65, 1995.
3. S. V. Mikhlyaev, "3D sensing of a nonstationary surface", Optical Engineering for Sensing and Nanotechnology
(ICOSN '99), Ichirou Yamaguchi; Ed., Proc. SPIE Vol. 3740, pp. 582-585, 1999.
z
P1 0' m1 xβ I0 α L1
Pβ x1
β x'1
L
U
z F U
x1 σ0
L d2 Ih
L2 D l1
n n1 z=h
α γ 2γ
L3 I1 σ1 σh
γ d1 l2=2hcos(α) Fig. 2. The geometry of
P h Fig. 1. The
m principle of α z=0 a probing Gaussian
L1 x beam reflected from a
P0 triangulation
0
measurements. l1=h/cos(α) surface z=h.
.
7K ∆ K PP α 
α  σ  0  

 0
 α  σ 
σ0 0  β
 0
 

 0
α  σ 
σ0  β
 0

 α  σ 
 σ0 

σ0 
 










K PP K PP

                 

Fig. 3. The error ratio T ( h ) = ∆h1 / ∆hβ for different values of parameters. Fig. 4. The absolute values of measurement errors.

Proc. SPIE Vol. 4416 403

Downloaded From: http://proceedings.spiedigitallibrary.org/ on 07/03/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx