sensors

Article
Optical Inspection System for Gear Tooth Surfaces
Using a Projection Moiré Method
Yi-Cheng Chen * and Jr-Yi Chen
Department of Mechanical Engineering, National Central University; Taoyuan 32001, Taiwan;
circle04617@gmail.com
* Correspondence: ethan@ncu.edu.tw




Received: 31 January 2019; Accepted: 21 March 2019; Published: 25 March 2019


Abstract: The demand for rapid online optical inspection of gear tooth surfaces is increasing,
especially for precision gears. In this study, a non-contact optical measurement method was
established for the inspection of gear tooth surfaces. For the system architecture, a halogen lamp was
selected as the light source, and a collimated beam was produced by an autocollimator. Subsequently,
moiré fringes were formed as the collimated beam went through the two linear gratings. The moiré
fringes projected on the gear tooth surface were recorded with a charge-coupled device (CCD) camera,
and the contour of the gear tooth surface was estimated and reconstructed from the phase information
of the fringes by our developed computer codes. To verify the accuracy of the system, a spur gear
tooth surface measured by a commercial coordinate measuring machine (CMM) was defined as the
reference tooth profile. The tooth topography, involute profile deviation, and axial-direction deviation
were successfully calculated by measuring the deviation of the optically measured surface based on
the reference gear tooth profiles measured using the CMM.

Keywords: optical inspection; projection moiré; gear tooth topography

1. Introduction
Gears are crucial components in the power transmission of automatic robotics and transportation
vehicles. With the increasing demand for high-precision gears for speed reducers used in industrial
robots and electric vehicles, high efficiency, as well as accurate in-situ inspection of the gear profile
and quality grade, are essential in the gear production process. Moreover, the resolution required for
high-quality gear profile inspection is in the range of a few micrometers.
Conventionally, the gears are measured by contacting probes on the tooth surface with multiple
points or by scanning several traces on the tooth surface. The industrial measurement facility for gear
tooth surfaces is usually a coordinate measuring machine (CMM) with contact probing. The resolution
increases with the increase of contact points on the tooth surface. However, the overall measuring time
for a gear workpiece also increases with an increase in the number of measuring points. Therefore,
the common practice is to measure only a few teeth, typically four teeth oriented 90 degrees apart.
Moreover, only a few traces on one tooth surface are measured, including the involute profile and
axial-direction traces. Consequently, these measurements present only partial information of all the
gear teeth.
Recently, optical metrology has been applied in measuring 3-D surfaces, including gear tooth
surfaces. Optical metrology possesses the advantages of being a non-contact method and having high
resolution and efficiency for workpiece contour measurement and surface quality inspection. Several
methods exist for measuring optical topography, including the stereoscopic method, structured light,
laser triangulation, conoscopic holography, the moiré method, and heterodyne interferometry [1–13].

Sensors 2019, 19, 1450; doi:10.3390/s19061450 www.mdpi.com/journal/sensors

Sensors 2019, 19, 1450 2 of 16

The structured light method is becoming popular for handheld scanning devices that have
relatively low precision but are more convenient. Peters et al. [1] applied structured light patterns
for the high-efficiency measurement of gear tooth surfaces. A charge-coupled device (CCD) camera
was used to capture the deformed stripes produced by the interaction between the light pattern
and the tooth surface. Their measuring area was a few square centimeters, and the resolution was
approximately 1 µm.
Several different interferometric methods have been applied for 3-D contouring.
Laser interferometry has high resolution but is sensitive to vibration and air turbulence during the
phase-shifting process. Fang et al. [2] used laser interferometry to compensate for the inevitable
installation errors that occur when measuring the gear tooth surface. Their experimental results
demonstrated the feasibility of their compensation method.
The laser triangulation technique is the most common method for online non-contact inspection;
however, it has fundamental limitations of limited precision and short standoffs. In addition, laser
triangulation is used to measure 3-D contours, especially for large workpieces that conventional
CMM cannot measure. Smith et al. [3] used point laser triangulation for the inspection of helical
gear. They studied an optimal path plan for helical gear profile inspection with a point laser
triangulation probe to avoid collision and create a complex sensor-to-surface orientation. Recently,
Auerswald et al. [4] proposed a laser line triangulation sensor for a large cylindrical involute gear with
a pitch diameter of 922 mm and a face width of 246 mm. They developed a laser line triangulation
sensor that can provide 1280 points at a 25-mm line width, and the speed can reach 200 lines/s.
Their measurement deviation is ±8.2 µm due to random and systematic errors. The measuring speed
is fast, but the accuracy can be improved.
Conoscopic holography has the advantages of a high viewing angle, robustness, and a high
point sampling rate. It has also been used for 3-D profile measurement with CMM [5] and in adverse
environments [6]. Fernández et al. [5] integrated a conoscopic holography sensor onto a CMM and
obtained an accuracy similar to that of laser triangular sensors. The errors of the conoscopic holography
sensor compared with the CMM results range from 4 µm to 18 µm. Enguita et al. [6] developed an
optical profilometer for distance calculation based on conoscopic holography and used it in a harsh
steelmaking environment. The resolution was 0.2 mm with a long working distance of 1200 mm when
they used conoscopic holography to inspect hot (over 800 ◦ C) steel slabs. Álvarez et al. [7] reviewed
the application of conoscopic holography techniques for submicron roughness measurement and
high-temperature surface detection. They reported that multiple wavelength conoscopic holography
can correctly resolve 3-µm steeps, extending the measurement range of the conoscopic holography
to online roughness measurement and microdefect detection. Moreover, an online inspection system
using conoscopic holography integrated with a robot was implemented for inspecting automobile
synchronizer hubs. The 3-D data were acquired and measured automatically for every workpiece, and
the precision was reported to nearly ±1 µm.
The moiré technique, which is similar to interferometry, has been applied in several moiré methods
in 3-D surface contouring, including projection moiré, shadow moiré, and moiré with heterodyne
interferometry [8,9]. Chang et al. [10] proposed a method for reconstructing surface profiles based on
projection moiré and heterodyne interferometry. The projection grating moves at a constant velocity
along the grating plane to produce a series of sinusoidal waves, similar to heterodyne interferometric
signals. The resolution of this system is improved to approximately 1.9 µm by combining the projection
moiré and heterodyne interferometry.
Jeong and Kim [11] presented a high-speed moiré method by performing 3-D contouring with
one fringe captured on the basis of the color grating projection. Practically, the accuracy of the moiré
method can be affected by errors induced during phase-shifting. Mu et al. [12] investigated the error
analysis of the phase-shifting in shadow moiré by changing the incidence angle of the collimating
beams. Experiments were performed to verify the proposed technique, and the results were found to
be 5% better than those of the conventional phase-shifting technique.
Sensors 2019, 19, 1450 3 of 16

Sciammarella et al. [13] applied projection moiré in evaluating gear contours and reported the
feasibility of using the projection moiré method. They produced the topographic information on a gear
surface and compared it with that provided by a mechanical tactile device. The involute profiles of the
gear measured by the optical method and by the mechanical method were examined and compared.
However, the lead profile and 3-D topology errors, which are frequently presented in gear inspection
report sheets, were not studied in their research.
In addition to the academic research on the optical metrology of gears, the industry has focused
on developing commercial machines for the optical measurement of gear surfaces. Nikon announced
the launch of HN-C3030 for 3-D gear surface measurement [14]. Laser structured patterns are projected
on the gear tooth surface, and the computer numerical control (CNC) measurement machine has a
five-axis movement. The resolution for this non-contact gear surface-measuring machine is reported to
be 5 µm. Recently, Gleason announced the development of a gear rolling tester with laser inspection
on a single machine [15]. The optical method for the profile inspection of cylindrical gears is reported
to be ten times faster than the conventional probing method.
This study proposed an optical measurement method for the inspection of gear tooth surfaces.
First, projection moiré fringes were formed on the gear tooth surface. Subsequently, the moiré fringe
information produced by the five-step phase-shifting was used to determine the phase information
to reconstruct the surface topography, and the contour of the gear tooth was estimated using our
developed computer codes. To verify the accuracy of the inspection system, a spur gear tooth surface
measured by a commercial CMM was defined as the reference tooth profile. The 3-D tooth topography,
involute profile deviation, and lead deviation were successfully calculated by comparing the deviation
of the optically measured surface and the reference gear tooth profiles using the CMM. Moreover, a
gauge block was measured using our optical method to confirm measurement accuracy.

2. Projection Moiré Measurement System

2.1. Projection Moiré Fringes

Moiré fringes refer to a beat pattern formed by two gratings with approximately equal spacing.
The precision depends on the numbers of fringes used in the measurement. For phase measurement,
the surface height can be calculated as 1/100 of a fringe.
For two straight line gratings, G1 and G2 , the intensity transmission functions, IG1 (x,y) and IG2 (x,y),
can be expressed as follows [8,16]:
∞
IG1 ( x, y) = α1 + ∑ b1n cos[nϕ1 (x, y)], (1)
n =1

∞
IG2 ( x, y) = α2 + ∑ b2m cos[nϕ2 ( x, y)] (2)
m =1

where ϕ(x,y) is the function representing the basic shape of the grating lines; α1 and α2 are the average
intensities; b1n and b1m denote the grating line profiles, such as square, triangular, and sinusoidal
wares. Straight line gratings are used in this study; therefore, the above equations can be simplified
and rewritten as follows:
IG1 ( x, y) = α1 + b1 cos[ ϕ1 ( x, y)], (3)

IG2 ( x, y) = α2 + b2 cos[ ϕ2 ( x, y)] (4)

When the two gratings are superimposed, the resulting intensity transmission function can be
attained by the product of Equations (3) and (4):

IG1 IG2 = α1 α2 + α1 b2 cos[ ϕ2 ( x, y)] + α2 b1 cos[ ϕ1 ( x, y)]

(5)
+b1 b2 cos[ ϕ1 ( x, y)] cos[ ϕ2 ( x, y)],
Sensors 2019, 19, 1450 4 of 16

The first three terms of Equation (5) contain the information for the two patterns separately. The
interesting term is the last term of Equation (5), and it can be expressed and rewritten as follows:

b1 b2 cos[ ϕ1 ( x, y)] cos[ ϕ2 ( x, y)] = 12 b1 b2 cos[ ϕ1 ( x, y) + ϕ2 ( x, y)]

(6)
+ 21 b1 b2 cos[ ϕ1 ( x, y) − ϕ2 ( x, y)],

where cos[ ϕ1 ( x, y) + ϕ2 ( x, y)] is the high-frequency portion of the two superimposed gratings, whereas
cos[ ϕ1 ( x, y) − ϕ2 ( x, y)] is the low-frequency portion. Generally, the low-frequency term can be used
to predict the moiré fringes and is observable by human eyes and optical sensors.
If the two gratings form an angle of 2α, and the y-axis bisects the intersecting angle, the two
grating functions can be rewritten as follows:

2π
ϕ1 ( x, y) = ( x cos α + y sin α), (7)
λ1

2π
ϕ2 ( x, y) = ( x cos α − y sin α), (8)
λ2
where λ1 and λ2 denote the grating period of the two gratings. Equations (7) and (8) can be rewritten
as follows:
2π 4π
ϕ1 ( x, y) − ϕ2 ( x, y) = x cos α + y sin α, (9)
λb λ
where λ = (λ1 + λ2 ) is the average grating period, and λb = λ1 λ2 /(λ1 + λ2 ) is the beat wavelength
between the two gratings.
According to Equation (6), the moiré fringes’ patterns can be lines whose centers satisfy the
following condition:
ϕ1 ( x, y) − ϕ2 ( x, y) = 2Mπ, (10)

where M is an integer and represents the fringe order. Substituting Equation (9) into Equation (10)
provides the following expression:

2π 4π
x cos α + y sin α = 2Mπ, (11)
λb λ

Furthermore, in this study, the two gratings have different periods and are parallel to each other,
which leads to the second term of Equation (9) being zero. The moiré fringes will be lines that satisfy
the following equation:
λ λ2
Mλb = x = M( 1 ), (12)
λ1 − λ2
Equation (12) demonstrates that the formed fringes will be equal-spaced lines parallel to
the gratings lines. Consequently, the superimposition of two gratings’ lines is similar to the
superimposition of two plane waves. Bright fringes form when the two waves are in phase, whereas
dark fringes form when they are out of phase. The moiré fringes of two straight line gratings can
determine the center of interference for fringes formed by interference from two plane waves.

2.2. Five-Step Phase-Shifting

The phase-shifting technique was used for projection moiré fringes to attain quantitative height
information. The errors in the phase-shifting technique include errors in phase-shifting, environmental
disturbances, such as air turbulence and vibration, and the nonlinearity of the recorded intensities of
the CCD detector [17]. The Hariharan algorithm is less sensitive to phase-shifting calibration errors
compared with three-step and four-step phase-shifting algorithms.
The five-step phase-shifting method proposed by Hariharan et al. [17,18] is used in this study.
In this method, five measurements of the fringe intensities are conducted and recorded. The step
Sensors 2019, 19, 1450 5 of 16

for each phase-shift is π/2, and the phases corresponding to each step are −π, −π/2, 0, π/2, and
π. After the fringes are captured, the intensity on each pixel of the sensor for the five steps can be
expressed as follows:
I1 (i, j) = A(i, j) + B(i, j) cos[ ϕ(i, j) − π ] (13a)
π
I2 (i, j) = A(i, j) + B(i, j) cos[ ϕ(i, j) − ] (13b)
2
I3 (i, j) = A(i, j) + B(i, j) cos[ ϕ(i, j)] (13c)
π
I4 (i, j) = A(i, j) + B(i, j) cos[ ϕ(i, j) + ] (13d)
2
I5 (i, j) = A(i, j) + B(i, j) cos[ ϕ(i, j) + π ] (13e)

where (i,j) is the pixel coordinate, IK (i,j) represents the intensity at Kth phase shifting (K = 1~5), A(i,j)
is the constant term, that is, the intensity bias, B(i,j) is the intensity amplitude, and ϕ(i,j) is the phase
information to be determined at the measurement point.
After the five equations are solved (Equation (13a–e)) simultaneously, the phase at each
measurement point can be calculated as an arctangent of a function of the intensities measured
at the same measurement point at each phase shifting:

2( I2 − I4 )
ϕ(i, j) = tan−1 (14)
2I3 − I 5 − I 1

Phase unwrapping is subsequently applied to remove the discontinuous 2π phase jumps.

The unwrapped phase map of the measured profile is applied to determine the height quantitatively.
Figure 1 illustrates the projection moiré fringes on the gear tooth surface as viewed by a CCD
camera. As demonstrated by Figure 1, a reference plane is placed at the middle of the tooth height,
and an angle α is formed between the projection fringes and the viewing direction of the CCD camera.
The surface height h(x,y) measured relative to the reference plane is as follows:

2OB NPo
h( x, y) = 2OA = = = NPn (15)
tan α tan α
where Pn is the contour interval (the height between adjacent contour lines in the CCD viewing angle),
and Po is the spacing of the lines perpendicular to the CCD viewing direction. The fringe order N is
the following:
ϕ( x, y)
N= . (16)
2π
Moreover, the larger the angle α is, the smaller the contour spacing is. Theoretically, the sensitivity
is maximized when angle α is 90◦ . By contrast, the measurement sensitivity approaches zero when
angle α is near 0◦ . This case is not applicable in practice. However, a general guideline is that the
angle chosen should be no larger than the maximum slope of the measured surface. In addition,
shadows that lead to areas with missing data should be avoided, because the shadowed area cannot be
contoured [17].
Sensors 2018, 18, x FOR PEER REVIEW 6 of 16

Sensors2018,
Sensors 19,x1450
2019,18, FOR PEER REVIEW 66ofof1616

Figure 1. Illustration of the angle formed by the charge-coupled device (CCD) viewing direction and
Figure
the 1. Illustration
projection of theand
moiré fringe, angle formed by
a reference the charge-coupled
plane device (CCD) viewing direction and
for the height estimation.
Figure 1. Illustration of the angle formed by the charge-coupled device (CCD) viewing direction and
the projection moiré fringe, and a reference plane for the height estimation.
the projection moiré fringe, and a reference plane for the height estimation.
2.3. Experimental Framework
2.3. Experimental Framework
Figure 2 illustrates
2.3. Experimental Frameworkthe framework of the proposed projection moiré technique established in this
Figure 2 illustrates the framework of the proposed projection moiré technique established in
study, and Figure 3 displays the setup of the optical measurement system on the optical bench. As
this Figure
study, 2and
illustrates
Figure 3the framework
displays of theof
the setup proposed
the opticalprojection moiré technique
measurement system onestablished
the opticalin this
bench.
exhibited in Figure 2, a halogen lamp was used as the illumination source, and the collimated beam
study, and Figure 3 displays the setup of the optical measurement system on
As exhibited in Figure 2, a halogen lamp was used as the illumination source, and the collimated the optical bench. As
was produced by a collimator. As Figures 2 and 3 illustrate, two projection lenses were used to
exhibited
beam wasinproduced
Figure 2, bya halogen lamp As
a collimator. was used as
Figures the illumination
2 and source,
3 illustrate, two and the
projection collimated
lenses beamto
were used
enlarge the projected fringe period from grating 1, to form the moiré fringes with fringes projected
was produced by a collimator. As Figures 2 and 3 illustrate, two projection lenses
enlarge the projected fringe period from grating 1, to form the moiré fringes with fringes projected were used to
from grating 2 on the tooth surface. Subsequently, the moiré fringes were built up on the gear tooth
enlarge the projected fringe period from grating 1, to form the moiré fringes with
from grating 2 on the tooth surface. Subsequently, the moiré fringes were built up on the gear tooth fringes projected
surface as the collimated beam passed through the two linear gratings. A linear stage driven by a
from grating
surface as the2 on the toothbeam
collimated surface. Subsequently,
passed through the thetwomoiré fringes
linear were built
gratings. up on
A linear thedriven
stage gear tooth
by a
servo motor enabled us to move grating 1 in the five-step phase-shifting. A CCD camera was used to
surface
servo motor enabled us to move grating 1 in the five-step phase-shifting. A CCD camera was used ato
as the collimated beam passed through the two linear gratings. A linear stage driven by
capture the images of moiré fringes formed on the gear tooth surface. Table 1 summarizes the major
servo motor
capture the enabled
images of usmoiré
to move grating
fringes 1 in the
formed five-step
on the phase-shifting.
gear tooth A CCD
surface. Table camera wasthe
1 summarizes used to
major
components of the projection moiré system used in this study.
capture the images
components of the of moiré fringes
projection moiréformed
system on theingear
used thistooth
study.surface. Table 1 summarizes the major
components of the projection moiré system used in this study.

Figure 2. Illustration of the experimental framework of the projection moiré system.

Figure 2. Illustration of the experimental framework of the projection moiré system.

Figure 2. Illustration of the experimental framework of the projection moiré system.

 Sensors 2019, 19, 1450 7 of 16
Sensors 2018, 18, x FOR PEER REVIEW 7 of 16

Figure 3. Photo of the experimental setup of the projection moiré system.

Figure 3. Photo of the experimental setup of the projection moiré system.
Table 1. Components of the measurement system.
Table 1. Components of the measurement system.
Component Specification
Component
Illumination Specification
Halogen source
Illumination
Collimator Halogen source
Trioptics 3-100-093
Projection module Collimator TriopticsPitch:
3-100-093
Gratings (×2) 200 µm
Projection module Gratings (×2) Pitch: 200 μm
f = 75 mm
Biconvex lens (×2) f = 75 mm
Biconvex lens (×2) f = 100 mm
f = 100 mm
Stepping servo motor and driver Oriental motor AR24SAK-1
Stepping servo motor and driver Oriental motor AR24SAK-1
AFM-40-C5
AFM-40-C5
Phase-shifting
Phase-shifting module
module LinearLinear
stagestage Repeatability: ±2 µm
Repeatability: ±2 μm
Control
Control systemsystem LABVIEW LABVIEWcodescodes
CCD CCD sensor
sensor Basler Basler acA640-90gm
acA640-90gm
ImageImage
capture module
capture module CameraCamera Fujinon Fujinon
ImageImage
processing
processing MATLAB MATLABcodescodes

Figure 4a illustrates
Figure the measured
4a illustrates the measuredspur spur
gear gear
(KHK SS2-18),
(KHK and its
SS2-18), andmajor design
its major parameters
design are are
parameters
listedlisted
in Table 2. The
in Table 2. module
The module of the
of gear is 2 mm,
the gear its face
is 2 mm, width
its face is 20ismm,
width 20 mm,andand
thethe
whole tooth
whole depth
tooth depth is
is 2.25
2.25 times that of the module (i.e., 4.5 mm). Thus, the area of one side of the tooth flank to be to
times that of the module (i.e., 4.5 mm). Thus, the area of one side of the tooth flank be
measured
measured is approximately 20 mm by 2.25 mm. Figure 4b represents the projected
is approximately 20 mm by 2.25 mm. Figure 4b represents the projected moiré fringes on the gear tooth moiré fringes on
the gear toothThe
surface. surface.
period The
of period of the
the moiré moiré
fringes fringes displayed
displayed in Figure in
4bFigure 4b isThe
is 806 µm. 806 projected
μm. The projected
moiré fringes
moiréare fringes are presented on three adjacent tooth surfaces, which implies
presented on three adjacent tooth surfaces, which implies that the optical method that the optical
may method
inspect the
may three
inspect the three tooth surfaces in one measurement. Moreover, the incident
tooth surfaces in one measurement. Moreover, the incident angles of the projected fringes angles of the
on the
projected fringes on the gear tooth surface and the viewing angle of the CCD
gear tooth surface and the viewing angle of the CCD camera have both been adjusted and tuned to camera have both been
adjusted and tuned
minimize to minimize
the erroneous phasethe erroneousoccurring
information phase information
on the tooth occurring on theinspection.
surface under tooth surface
Besides,
under theinspection. Besides,for
region of interest theinspection
region of interest
is only onefor inspection is only
tooth surface onestudy.
in this tooth surface in this study.
Sensors 2019, 19, 1450 8 of 16
Sensors 2018, 18, x FOR PEER REVIEW 8 of 16

(a) (b)
Figure 4. (a) Photo of the 18-tooth spur gear; (b) projected moiré fringes on the gear tooth surface.
surface.

Table 2. Design parameters of the measured spur gear.

Table 2. Design parameters of the measured spur gear.
Parameters
Parameters Values
Values
Module Module
(mm)(mm) 22
Teeth Number 18
Teeth Number 18
Face Width (mm) 20
Face Width (mm)
Pressure Angle (degrees) 20
20
PressurePitch
Angle (degrees)
Diameter (mm) 20
36
Addendum
Pitch DiameterDiameter
(mm) (mm) 40
36
Addendum Diameter (mm) 40
2.4. GearTooth Surface Measurement Procedure
2.4. GearTooth Surface Measurement Procedure
The measurement process of the spur gear tooth surface using the projection moiré method is
The measurement
illustrated in Figure 5. Theprocess of the
process spur gearastooth
is described surface using the projection moiré method is
follows:
illustrated in Figure 5. The process is described as follows:
(1) Perform the distortion correction of the CCD camera by obtaining the distortion correction
(1). Perform the distortion correction of the CCD camera by obtaining the distortion correction
coefficients with a grid distortion target, as illustrated in Figure 6a. Perform depth calibration
coefficients with a grid distortion target, as illustrated in Figure 6a. Perform depth calibration to
to obtain the calibration coefficient of the magnification along the tooth depth, as illustrated in
obtain the calibration coefficient of the magnification along the tooth depth, as illustrated in
Figure 6b.
Figure 6b.
(2) Project the moiré fringes on the gear.
(2). Project the moiré fringes on the gear.
(3) Capture the deformed moiré fringes on the gear tooth surface.
(3). Capture the deformed moiré fringes on the gear tooth surface.
(4) Perform the five-step phase-shifting by moving grating 1 on a linear stage and record the images
(4). Perform the five-step phase-shifting by moving grating 1 on a linear stage and record the images
of moiré fringes at each phase-shifting step. Because the fringe pitch is 806 µm, the translation of
of moiré fringes at each phase-shifting step. Because the fringe pitch is 806 μm, the translation
the grating for each phase-shifting is 201.5 µm. Figure 7 presents five images of moiré fringes on
of the grating for each phase-shifting is 201.5 μm. Figure 7 presents five images of moiré fringes
the gear tooth surface for each phase-shifting step.
on the gear tooth surface for each phase-shifting step.
(5) Perform
(5). Perform the theimage
imageprocessing
processingand and phase
phase unwrapping
unwrapping of the
of the images
images of moiré
of moiré fringes
fringes usingusing
self-
developed Matlab codes. Phase unwrapping is conducted using the branch-cut method, andand
self-developed Matlab codes. Phase unwrapping is conducted using the branch-cut method, the
the noise can be filtered using a Butterworth low-pass
noise can be filtered using a Butterworth low-pass filter. filter.
(6) Reconstruct the
(6). Reconstruct the 3-D
3-D contour
contour ofof the
the gear
gear tooth
tooth surface
surface from
from the the calculated
calculated phase
phase information
information
according totothe
according themoiré
moiré fringes.
fringes. FigureFigure 8a shows
8a shows the original
the original phase onphase on the
the tooth toothobtained
surface surface
from the five-step phase-shifting, whereas Figure 8b displays the reconstructed 3-D gear tooth
surface.
 Sensors 2019, 19, 1450 9 of 16

obtained from the five-step phase-shifting, whereas Figure 8b displays the reconstructed 3-D gear
Sensors 2018, 18, x FOR PEER REVIEW 9 of 16
tooth surface.
Sensors 2018, 18, x FOR PEER REVIEW 9 of 16
(7) Compare and calculate the deviations of the optically measured surface from the CMM-measured
(7). Compare and calculate the deviations of the optically measured surface from the CMM-
reference surface.
(7).measured
Compare and calculate
reference surface.the deviations of the optically measured surface from the CMM-
measured reference surface.
ToTo compare
compare and
and verify
verify thethe measured
measured results
results from
from thethe optical
optical method,
method, thethe gear
gear tooth
tooth surface
surface
measured To
measured compare
byby and
a commercial verify
a commercial CMM the
CMM measured
is used
is used results
as as from
a reference
a reference the optical
surface.
surface. AA method,
photo
photo of of the
thethe
CMMgear
CMM tooth surface
measurement
measurement
measured
is presented byina commercial
Figure CMM is used as a reference surface. A7.6.5,
photo of the
the CMM measurement
is presented in Figure 9a.9a.
TheTheCMMCMM is aisBrown
a Brown && Sharpe
Sharpe XCEL
XCEL 7.6.5, andand accuracy
the accuracy of of
thethe
CMMCMM
is presented
is 3.0 + 4 in Figure
L/1000 µm, 9a. TheLCMM
where is a the
denotes Brown & Sharpe
length of the XCEL 7.6.5, feature.
measured and the accuracy
In of the
addition, CMM9b
Figure
is 3.0 + 4 L/1000 μm, where L denotes the length of the measured feature. In addition, Figure 9b
is 3.0 + 4 L/1000
illustrates μm, and
where L denotes the length theofmeasurement
the measuredpointsfeature. Intheaddition, Figure 9b
illustrates thethe radial
radial and axialaxial distribution
distribution of theofmeasurement points on theon gearsurface.
gear tooth tooth surface.
The
illustrates the
Themeasurement radial
total measurement and axial distribution
point number of
on one the measurement
gear points
toothissurface on
is 2460the gear
points. tooth
Curvesurface. The
total point number on one gear tooth surface 2460 points. Curve fitting wasfitting
appliedwas
total measurement point number on one gear tooth surface is 2460 points. Curve fitting was applied
to build the reference surface using the CMM for further calculation of the deviations from thethe
applied to build the reference surface using the CMM for further calculation of the deviations from
to build the
optically reference
measured surface
surface basedusing the reference
CMM forCMM-measured
further calculation of the deviations from the
optically measured surface based onon thethe
reference CMM-measured surface.
surface.
optically measured surface based on the reference CMM-measured surface.

Figure 5. 5.
Figure Process ofof
Process the proposed
the proposedoptical
opticalinspection
inspectionofofthe
thegear
geartooth
toothsurface.
surface.
Figure 5. Process of the proposed optical inspection of the gear tooth surface.

(a)(a) (b)
(b)
Figure 6. (a) Distortion coefficient correction with a grid distortion target; (b) calibration coefficient of
Figure
Figure 6. 6.
(a)(a) Distortion
Distortion coefficientcorrection
coefficient correctionwith
witha agrid
griddistortion
distortiontarget;
target;(b)
(b)calibration
calibrationcoefficient
coefficient
the magnification along the tooth depth.
of of
thethe magnification
magnification along
along the
the tooth
tooth depth.
depth.
 Sensors2019,
Sensors 2018,19,
18,1450
x FOR PEER REVIEW 10of
10 of16
16
Sensors 2018,2018,
Sensors 18, x18,
FOR PEER
x FOR REVIEW
PEER REVIEW 1010
ofof
1616

Figure 7. Moiré fringes on the gear tooth surface for each phase-shifting step.
Figure
Figure 7.7.Moiré
Moiré
7. Moiré
Figure fringes
fringes on
onon
fringes the
thethe gear
gear tooth
tooth
gear surface
surface forfor each
each phase-shifting
phase-shifting step.
step.
phase-shifting step.

(a) (b)
(b)
(a) (b)
Figure
Figure8.8.
Figure 8.(a)(a)
(a) Wrapped
Wrapped phase map
map map
phase on
on theon the
gear
the gear
tooth
gear toothand
surface
tooth surface and the
the region
surface and the
of region(b)
interest;
region ofofreconstructed
interest;(b)
interest; (b)
Figure 8. (a) Wrapped phase map on the gear tooth surface and the region of interest; (b)
reconstructed
3-D tooth surface.
reconstructed 3-D tooth surface.
reconstructed 3-D tooth surface.

(a)
(a) (b)
(a) (b)
(b)
Figure9.
Figure
Figure 9. (a)
9. (a) Tooth
(a) Tooth surface
surfacemeasured
Tooth surface measuredby
measured bybyaa commercial
acommercial coordinate
commercialcoordinatemeasuring
measuring
coordinate machine
machine
measuring (CMM);
(CMM);
machine (b)
(CMM);(b)
Figure 9. (a) Tooth
measuring pointssurface
on the measured
tooth by a commercial coordinate measuring machine (CMM); (b)
surface.
measuring
(b) measuring points on on
points thethe
tooth surface.
tooth surface.
measuring points on the tooth surface.
3.3.
3.Results
Results
Results
3. Results
Generally,
Generally,the
Generally, thedata
the data sheet
data sheet of
sheet of aaa gear
gear inspection
gear inspectionrepresents
inspection representsthe
represents thequality
the quality
quality ofofof the
the
the gear
gear
gear toothtooth
tooth surface
surface
surface inin
in Generally,
three
threeparts:
parts:the
(1)
(1)data sheetprofile
Involute
Involute of a gear
profile inspection
deviation;
deviation; (2)
(2) represents
lead
lead the quality of
(axial-direction)
(axial-direction) the gear and
deviation;
deviation; tooth
and surface
(3)
(3) in
topology
topology
three parts: (1) Involute profile deviation; (2) lead (axial-direction) deviation; and (3) topology
three parts: (1)
deviation.
deviation. TheInvolute profile
ideal profile deviation;
of the cross-section (2) lead (axial-direction)
of a gear tooth along the deviation; and (3)
radial direction topology
is an involute
deviation.The Theideal
idealprofile
profileofofthe
thecross-section
cross-sectionof of a gear tooth alongthe theradial
radialdirection
directionisisan aninvolute
involute
deviation. The ideal
curve. Typically,
curve. profile
the of the
profile cross-section
of the central of a gear
cross-section tooth
of theof along the
theblank radial
gear blank direction
is measuredis an involute
when the
curve.Typically,
Typically, thethe
profile of the
profile of central
the central cross-section
cross-section ofgear
the gearisblank measured when the
is measured involute
when the
curve.
profile Typically,
involute profilethe
deviation is profile is
deviation of represented,
represented, theascentral cross-section
as illustrated
illustrated in Figure of
in
10a.theIngear
Figure blank
10a. is measured
In addition,
addition, the thealong
profile when
profilethe the
along
lead
involute profile deviation is represented, as illustrated in Figure 10a. In addition, the profile along
involute
the lead
direction profile deviation
direction
is typically is represented,
is typically measured asalong
illustrated
the in Figure
axial of the10a.
direction ofIntheaddition,
toothon the
the profile
surface on the along
pitch
the lead direction is measured along the
typically measured axial
along direction
the axial direction tooth surface
of the tooth surface pitch
on circle,
the pitchas
thedepicted
lead direction
circle, asindepictedis
Figure typically
in Figure
10b. measured along
10b. Furthermore,
Furthermore, a the
topology axial
a topology
error direction of the
error represents
represents a 3-Dtooth surface
a 3-D topology
topology on the
deviation pitch
deviation
of the
circle, as depicted in Figure 10b. Furthermore, a topology error represents a 3-D topology deviation
circle, as depicted
of the
measured measured inreal
real tooth Figure
tooth10b. Furthermore,
surface afrom a tooth
a referencetopologytooth error inrepresents
surface a 3-D
in aofsetgrid
of grid topology
points. Indeviation
this study,
of the measured realsurface from
tooth surface reference
from a reference surface
tooth surfacea setin a set of points. In this
grid points. In study, the
this study,
of determination
the
themeasured real
determination
of thetooth
of thesurface
three from
types a
of reference
deviations tooth
was surface
also in
developed a set of grid
using points.
Matlab In
codes. this study,
the determination ofthree typestypes
the three of deviations
of deviations was was also also
developed
developed usingusing
Matlab codes.
Matlab codes.
the determination of the three types of deviations was also developed using Matlab codes.
 Sensors2019,
Sensors 2018,19,
18,1450
x FOR PEER REVIEW 11of
11 of16
16
Sensors 2018, 18, x FOR PEER REVIEW 11 of 16

(a) (b)
(a) (b)
Figure 10. (a) Illustration of the involute profile measurement at the central cross-section; (b)
Figure 10. (a) Illustration of the involute profile measurement at the central cross-section; (b) illustration
Figure 10. (a) ofIllustration
illustration of the
the lead (axial) involute
profile profile measurement
measurement at the central cross-section; (b)
on the pitch circle.
of the lead (axial) profile measurement on the pitch circle.
illustration of the lead (axial) profile measurement on the pitch circle.
Inthis
In thissection,
section, the
thetooth
toothsurface
surface of ofthe
thespur
spurgeargearmeasured
measuredby bythetheproposed
proposedopticalopticalmethod
methodisis
In this
comparedwith section,
withthe the
theCMM tooth surface
CMMresults,
results, of
and the spur gear measured by the proposed optical method is
compared and thethe results
results of ofthethe three
three typestypes of gear
of gear toothtooth surface
surface deviations
deviations are
compared with the
are calculated CMM results, thereafter.
and the results of the three types of gear tooth surface deviations
calculated and and represented
represented thereafter.
are calculated and represented thereafter.
3.1. Involute
3.1. InvoluteProfile
Profile
3.1. Involute Profile
Figure11a
Figure 11adepicts
depictsthetheradial
radialtooth
toothprofile
profileat atthe
thecentral
centralcross-section
cross-sectionof ofthe
thegear
gearblank
blankmeasured
measured
Figure
by the
by 11a
the optical depicts
optical projectionthe radial
projection moiré tooth
moiré method profile
method and at
and by the
by the central cross-section
the commercial
commercial CMM. CMM. Theof the gear
The deviation blank
deviation of measured
of the
the profile
profile
bymeasured
the opticalby
measured projection
by the moiré
optical method
method based and on by
thetheCMMcommercial
reference CMM.
was
the optical method based on the CMM reference was calculated and is illustrated inThe deviation
calculated and of
is the profile
illustrated in
measured
Figure by
11b. the
As optical method
demonstrated in based
Figure on the
11b, CMM
the mean reference
value was
of the
Figure 11b. As demonstrated in Figure 11b, the mean value of the involute profile deviation is calculated
involute and
profile is illustrated
deviation isin2.67
Figure
μm µm
2.67 11b.and
and As maximum
the demonstrated
the maximum in
deviationFigure
deviation 11b,
is 4.49 μm.
is 4.49theµm.
mean
The value
average
The of the
deviation
average involute
of theofprofile
deviation optical deviation
measurement
the optical is 2.67
measurement from
μm and
the CMMthe maximum
measurement deviation
is withinis 4.49
the μm. The
accuracy average
of the deviation
CMM (i.e.,
from the CMM measurement is within the accuracy of the CMM (i.e., within 3 µm), which implies of the
within optical
3 μm), measurement
which implies from
that the
thethat
CMM
proposed measurement is
optical optical
the proposed within
measurement the
measurement accuracy
method of the
leads leads
method CMM (i.e.,
to reasonably within
to reasonably 3
accurate μm), which
results
accurate implies
compared
results comparedthat the
withwith the
proposed
commercial optical
CMM. measurement
Furthermore, method
the leads
optical to
method reasonably
can accurate
inspect
the commercial CMM. Furthermore, the optical method can inspect and measure the whole tooth and results
measure compared
the whole with
tooth the
surface
commercial
surface in CMM.
in one measurement, Furthermore,
whereas
one measurement, the
the optical
whereas CMM method
the requires
CMM can inspect
scanning
requires orand
scanning measure
contacting the
or contacting thetooth
whole tooth
the surface
tooth surface
in a point-
surface in a
in point-by-point
one measurement,
by-point approach. whereas
Moreover,
approach. the
Moreover, CMM requires
it is possible scanning
to contour
it is possible or contacting
the surface
to contour the surface the
of several tooth surface
adjacent
of several in a point-
teethteeth
adjacent fromfromone
by-point approach.
measurement of Moreover,
the optical it is possible
method. to contour
Therefore, the
the surface
proposed of
one measurement of the optical method. Therefore, the proposed projection moiré method exhibits several
projection adjacent
moiré teeth
method from one
exhibits
measurement
comparable of
comparableaccuracy the optical
accuracy whilemethod.
while savingtime
saving Therefore,
time inthe
in the proposed
themeasurement
measurement projection
ofofgear
geartooth moiré
toothsurfaces.
surfaces. method exhibits
comparable accuracy while saving time in the measurement of gear tooth surfaces.

(a) (b)
(a) (b)
Figure 11. (a) Radial (involute) profiles measured by the CMM and optical system; (b) deviation of the
Figure 11. (a) Radial (involute) profiles measured by the CMM and optical system; (b) deviation of
involute profile of the projection moiré system from the CMM method.
Figure 11. (a) Radial
the involute profile(involute) profiles moiré
of the projection measured by from
system the CMM and optical
the CMM method. system; (b) deviation of
the involute profile of the projection moiré system from the CMM method.
3.2. Lead Profile
3.2. Lead Profile
Sensors 2019, 19, 1450 12 of 16

3.2. Lead Profile

Sensors 2018, 18, x FOR PEER REVIEW 12 of 16
Sensors 2018, 18, x FOR PEER REVIEW 12 of 16
The profiles along the lead (axial) direction on the pitch circle measured by the CMM and the
The profiles along the lead (axial) direction on the pitch circle measured by the CMM and the
optical method are presented
The profiles in Figure 12a. The deviation of circle
the lead profile measured by the optical
optical methodalong the lead
are presented (axial)
in Figure direction
12a. Theon the pitch
deviation measured
of the lead profile by the CMM
measured and
by the the
optical
method based
optical
method on the
method
based CMM
areonpresented
the CMMreference
Figurewas
inreference 12a. calculated
wasThe deviation
calculated and isisdepicted
of the
and in measured
lead profile
depicted in Figure12b.
Figure 12b.
byAsAs
the Figure
optical
Figure 12b12b
method
reveals, the based
maximumon the CMM reference
deviation along was
thecalculated
lead and
directionis depicted
is 3.63 in
µm, Figure
and
reveals, the maximum deviation along the lead direction is 3.63 μm, and the mean value of the 12b.
the As
meanFigure
value 12b of the
reveals,
deviation the
along maximum
the lead deviation
direction along
is 2.02 the
µm. lead
The direction
mean is
value 3.63
of μm,
the and the
deviation mean
is value
within
deviation along the lead direction is 2.02 μm. The mean value of the deviation is within the accuracy of
the the
accuracy
deviation
obtainedobtainedalong
by the the CMM.
byCMM.
the lead direction
Therefore,
Therefore,isthe
2.02 μm.
tooth
the The
tooth meanmeasured
surface
surface value of theby
measured deviation
by is within
the projection
the projection the accuracy
moiré
moiré method
method is is
obtained by
trustworthy the CMM. Therefore,
and efficient the tooth surface measured by the projection moiré method is
trustworthy and efficient and and is comparable
is comparable withwith
thethe resultsobtained
results obtained using
using aacommercial
commercial CMM.
CMM.
trustworthy and efficient and is comparable with the results obtained using a commercial CMM.

(a) (b)
(a) (b)
12. (a)12.
Figure Figure Lead profiles
(a) Lead measured
profiles by the
measured CMM
by the CMM and
andoptical
opticalsystem; (b)deviation
system; (b) deviationofofthethe lead
lead profile
profile
Figure
measured 12.
by (a)
the Lead profiles
projection measured
moiré by
method the CMM
from and
the optical
CMM system;
method.
measured by the projection moiré method from the CMM method. (b) deviation of the lead profile
measured by the projection moiré method from the CMM method.
3.3. Tooth
3.3.Surface Topology
Tooth Surface Topology
3.3. Tooth Surface Topology
FigureFigure
13a illustrates the 3-D
13a illustrates geargear
the 3-D tooth profile
tooth profilemeasured
measuredby by the projectionmoiré
the projection moiré method
method (black
(black
line) and Figure
the
line) CMM
and 13a illustrates
the CMM (redthe
(red line). 3-DThe
The
line). gear
3-D tooth
profiles
3-D profile
profiles from measured
from the
thetwo by the projection
twomethods
methods overlap
overlap moiré method
because
because thethe (black
deviations
deviations
line)
areand the CMM (red line). The 3-D profiles from the
are only aonly
few amicrometers.
few micrometers. TheThe gear
gear toothsurface
tooth surface is two
is methods
divided
divided into overlap
10
into × 5×
10 5because
grids the
whenwhen
grids deviationsthe
representing
representing
aretopology
only a few micrometers.
deviation. When The
the gear tooth
topology surface
deviation is divided
is into
calculated, 10 ×
the5 grids
tooth when
surface representing
defined thethe
from
the topology deviation. When the topology deviation is calculated, the tooth surface defined from
topology
CMM deviation. When
measurementisisconsideredthe topology
considered as deviation
as the is
the reference calculated,
reference surface. the tooth surface
Subsequently, defined
thethe
normal from the
(shortest)
the CMM
CMM measurement
measurement surface. Subsequently, normal (shortest)
distance from each is considered
grid point on the as reference
the referencesurfacesurface. Subsequently,
to the surface measured thebynormal (shortest)
the optical method
distance from
distance each grid point on the reference surfacetotothe the surface measured byoptical
the optical method
is calculated and recorded as the value of deviation at that grid point. Accordingly, themethod
from each grid point on the reference surface surface measured by the topology
is calculated andand
is calculated recorded
recorded as thethevalue of
of deviation atatthat
thatgrid
grid point. Accordingly, the topology
deviation of the surfaceasmeasured value by deviation
the projection moiré point.
method Accordingly,
from the CMM the topology
results is
deviation of
deviation theof surface
the measured
surface measured by the
by projection
the moiré
projection method
moiré from
method the
from
represented in Figure 13b. The mean value of the topology deviation is 2.81 μm, and the maximum CMMthe results
CMM is represented
results is
represented
in Figure 13b. The
deviation in Figure
mean
is 6.17 13b. The mean value of the topology deviation is 2.81
μm.value of the topology deviation is 2.81 µm, and the maximum deviation isμm, and the maximum
6.17 deviation
µm. is 6.17 μm.

(a)
(a) (b)
(b)

Figure 13. Spur gear tooth surface: (a) Topology of the gear tooth surface in 10 × 5 grids; (b) deviation
of the tooth surface topology measured by the projection moiré system from the CMM method.
Sensors 2018, 18, x FOR PEER REVIEW 13 of 16

Sensors 2019, 19, 1450 13 of 16

Figure 13. Spur gear tooth surface: (a) Topology of the gear tooth surface in 10 × 5 grids; (b) deviation
of the tooth surface topology measured by the projection moiré system from the CMM method.
4. Discussion
4. Discussion
The proposed optical method can inspect the whole tooth surface of the spur gear, and the three
types The proposedcommonly
of deviations optical method
used can inspect
in the the whole tooth
gear inspection reports surface of the spur
are calculated andgear, and theThree
presented. three
types of deviations commonly used in the gear inspection reports are calculated and
forms of deviations of the gear tooth surface measured by the projection moiré method from the CMM presented. Three
forms of
results aredeviations
illustratedof inthe gear 11–13.
Figures tooth surface
Table 3 measured
summarizesby the
themaximum,
projection moiré method
minimum, meanfrom the
value,
CMM results are illustrated in Figures 11–13. Table 3 summarizes the maximum,
and standard deviation of the three tooth surface deviations. The mean values of the tooth surface minimum, mean
value, and
deviation of standard deviation
the three forms of smaller
are all the threethantooth surface
3 µm, whichdeviations. The range
is the accuracy meanofvalues of the
the CMM tooth
used in
this study. In addition, the standard deviations of the three types of deviations are all in the rangethe
surface deviation of the three forms are all smaller than 3 μm, which is the accuracy range of of
CMM
1–2 µm.used in this study. In addition, the standard deviations of the three types of deviations are all
in the range of 1–2 μm.
Table 3. Summary of involute, lead, and topology deviations compared with the CMM results.
Table 3. Summary of involute, lead, and topology deviations compared with the CMM results.
Measured Tooth Surface Deviation (µm)
Measured Tooth Surface Deviation (μm)
Mean Maximum Minimum Standard Deviation
Mean Maximum Minimum Standard Deviation
Involute profile 2.67 4.49 2.59 × 10 − 2 1.44
Involute profile 2.67 4.49 2.59 × 10−2 1.44
Lead profile
Lead profile 2.02 2.02 3.63
3.63 0.270.27 1.24 1.24
3-D Topology
3-D Topology2.81 2.81 6.17
6.17 0.410.41 1.31 1.31

According to
According to the experimental
experimental results
resultspresented
presentedininTable
Table3,3,the
thedeviations
deviations of of
thethe
three forms
three are
forms
all smaller than 3 μm, which is the accuracy range of the CMM. To determine
are all smaller than 3 µm, which is the accuracy range of the CMM. To determine the accuracy of the accuracy of our
proposed
our proposedmethod,
method, a Mitutoyo
a Mitutoyo grade
gradeKKgauge gaugeblock
blockof of dimension 30(L) × × 8(W) × × 3(H) mm mm33 was
was
measured, and the deviation was calculated. Figure 14a,b depicts the moiré fringes
measured, and the deviation was calculated. Figure 14a,b depicts the moiré fringes on the gage block on the gage block
andthe
and thereconstructed
reconstructed 3-D3-D profile
profile by ourbyoptical
our optical
method,method, respectively.
respectively. In addition,
In addition, Figure 14cFigure 14c
illustrates
illustrates
the the height reconstructed
height reconstructed by our method by and
our themethod andheight
nominal the nominal height
of the gage of the
block. Thegage block.
average The
value
average
of value
the height of the height
deviation is 2.76deviation
µm. is 2.76 μm.
Therefore,the
Therefore, theaccuracy
accuracyof ofour
ouroptical
opticalmeasurement
measurementmethod
methodisis2.76
2.76µm,
μm,which
whichisisjustified
justifiedbybythe
the
gauge block
gauge block measurement
measurement and is also also consistent
consistentwith
withcomparisons
comparisonsofofthe theCMM
CMMand andoptical
optical results in
results
the preceding section. The accuracy of our system is comparable with the
in the preceding section. The accuracy of our system is comparable with the reported accuracy ofreported accuracy of 0.6–
2.6 μm,µm,
0.6–2.6 andandthethe
average
averagedifference
difference ofof1.61.6μm
µmwas
wasconfirmed
confirmedby byusing
using aa Zeiss CMM and and following
following
the method in [7]. The sources of errors in our system are
the method in [7]. The sources of errors in our system are discussed in discussed in the next paragraph. Moreover,
paragraph. Moreover,
theaccuracy
the accuracyof ofour
oursystem
systemcancanbebefurther
furtherimproved
improvedby bymodifying
modifyingour ourprojection
projectionmoiré
moirésystem.
system.

(a) (b)
Figure 14. Cont.
Sensors 2019, 19, 1450 14 of 16
Sensors 2018, 18, x FOR PEER REVIEW

(c)
Figure 14. Gauge block measurements: (a) moiré fringes on the gauge block; (b) reconstructed 3-D
Figure 14. Gauge block measurements: (a) moiré fringes on the gauge block; (b) reconstructed
profile; (c) optically measured profile and gauge block profile.
profile; (c) optically measured profile and gauge block profile.
The errors during the phase-shifting measurement may come from the following sources:
(1) The data acquisition system; The(2)errors during thedisturbance;
environmental phase-shifting measurement
(3) the nonlinearitymay come
of the from the following sou
sensors;
(4) quantizing the signal from analog to digital; (5) source instability; (6) stray reflection. The nonlinearity
The data acquisition system; (2) environmental disturbance; (3) the data of the sen
quantizing the signal from analog to digital; (5) source instability;
acquisition system includes the camera and the phase-shifting mechanism. A five-step phase-shifting (6) stray reflection. T
algorithm was used in this acquisition
study to system includes
minimize the camera
the effect and the phase-shifting
of phase-shifting mechanism.
errors. Moreover, dial A five-step phase
algorithm was used in this study to minimize the effect of phase-shifting errors. Moreo
indicators and laser displacement sensors were applied to improve the accuracy of each phase-shifting
indicators and laser displacement sensors were applied to improve the accuracy of eac
step. The accuracy can be further enhanced by using a more accurate phase-shifting motorized stage,
shifting step. The accuracy can be further enhanced by using a more accurate phase
such as a piezoelectric actuator.
motorized stage, such as a piezoelectric actuator.
Although the magnification calibration along the depth, as well as the distortion calibration of the
Although the magnification calibration along the depth, as well as the distortion calib
camera, were performed for the system, the accuracy can be further improved if a telecentric camera
the camera, were performed for the system, the accuracy can be further improved if a te
lens system is used for image capturing. In addition, the vibration and air disturbance were kept
camera lens system is used for image capturing. In addition, the vibration and air disturban
to a minimum because the experiments were performed on a vibration-isolated optical bench in a
kept to a minimum because the experiments were performed on a vibration-isolated optical
laboratory with shielding curtains. Surface reflection and shadows should be avoided; otherwise, these
a laboratory with shielding curtains. Surface reflection and shadows should be avoided; ot
may result in the saturation of sensor pixels and missing data on the corresponding measured area.
these may result in the saturation of sensor pixels and missing data on the corresponding m
In summary, the proposed projection moiré method is considered a trustworthy and accurate
area.
non-contact method for gear tooth surface inspection. Moreover, the precision is related to the number
In summary, the proposed projection moiré method is considered a trustworthy and
of fringes used in the measurement; it can be further improved by applying different gratings to
non-contact method for gear tooth surface inspection. Moreover, the precision is relate
the projection moiré system. It will also be interesting and challenging to inspect two tooth surfaces
number of fringes used in the measurement; it can be further improved by applying different
simultaneously in the future, which may be affected by the shadowed area, reflection, and the steep
to the projection moiré system. It will also be interesting and challenging to inspect two tooth
slope of the tooth surface. The advantages of the projection moiré method are that it is contact-free,
simultaneously in the future, which may be affected by the shadowed area, reflection, and t
accurate, allows time savings, and that the tooth surface can be measured in one measurement.
slope of the tooth surface. The advantages of the projection moiré method are that it is con
The optical method is expected to be applied to other 3-D surfaces as well, such as ball-screws, cams,
accurate, allows time savings, and that the tooth surface can be measured in one measurem
screw compressors, and turbine blades.
optical method is expected to be applied to other 3-D surfaces as well, such as ball-screw
5. Conclusions screw compressors, and turbine blades.

An optical inspection 5.method for gear tooth surfaces using the projection moiré is proposed and
Conclusions
studied in this paper. The tooth surface of a spur gear was measured using the projection moiré method,
An opticalwith
and the measured results were compared inspection method
those from for gear tooth
a commercial CMM. surfaces usingthree
In addition, the projection
forms moiré is propo
of tooth surface deviations,studied
namelyininvolute
this paper. Thelead
profile, tooth surface
profile, andof3-D
a spur gear measured
topology, was measuredby theusing the projectio
method, and the measured results were compared with those from
optical method from the reference surface (CMM results) and presented in gear inspection sheets, were a commercial CMM. In a
calculated and illustrated.three forms revealed
The results of tooth that
surface deviations,
the mean valuesnamely involute
of the profile profile, lead
deviations of theprofile, and 3-D t
tooth surface measured by the proposed projection moiré method were within 3 µm, which wasresults)
measured by the optical method from the reference surface (CMM the and presented
inspection sheets, were calculated and illustrated. The results revealed that the mean valu
profile deviations of the tooth surface measured by the proposed projection moiré meth
Sensors 2019, 19, 1450 15 of 16

accuracy range of the CMM used in this study. The accuracy of our optical measurement method is
2.76 µm, which was justified using the gauge block measurement.
The results imply that the projection moiré method is an alternative, trustworthy, rapid, and
non-contact method for measuring gear tooth surfaces. Furthermore, the precision grades of the
measured gear, such as DIN (Deutsches Institut für Normung), ISO (International Organization for
Standardization), and AGMA (American Gear Manufacturers Association) grades, can be determined
and classified using a theoretical tooth surface geometry as the reference surface when calculating the
involute, lead, and topology deviations.

Author Contributions: Conceptualization and methodology: J.-Y.C. and Y.-C.C.; experiment, Matlab, and Labview
codes: J.-Y.C.; writing—original draft preparation: J.-Y.C. and Y.-C.C.; writing—review and editing: Y.-C.C.;
supervision and project administration: Y.-C.C.
Funding: This research was funded by the Ministry of Science and Technology, Taiwan,
grant MOST-107-2221-E-008-047-MY2.
Acknowledgments: The authors would like to acknowledge Ju-Yi Lee for his support related to technical
discussions and materials in building the projection moiré system, Chien-Sheng Liu for supporting motorized
stage for phase-shifting, and Chien-Yau Huang at the Taiwan Instrument Technology Research Center for assisting
in the CMM measurement.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Peters, J.; Goch, G.; Gunther, A. Helical gear measurement using structured light. In Proceedings of the XVI
IMEKO World Congress, Wien, Austria, 25–28 September 2000; pp. 227–230.
2. Fang, S.; Zhu, X.; Yang, P.; Cai, Q.; Komori, M.; Kubo, A. Analysis and compensation method for installation
error in measuring gear tooth flank with laser interferometry. Opt. Eng. 2014, 53. [CrossRef]
3. Smith, K.B.; Zheng, Y.F. Optimal path planning for helical gear profile inspection with point laser
triangulation probes. J. Manuf. Sci. Eng. Trans. ASME 2001, 123, 90–98. [CrossRef]
4. Auerswald, M.M.; von Freyberg, A.; Fischer, A. Laser line triangulation for fast 3D measurements on large
gears. Int. J. Adv. Manuf. Technol. 2019, 100, 2423–2433. [CrossRef]
5. Fernández, P.; Blanco, D.; Valiño, G.; Hoang, H.; Suárez, L.; Mateos, S. Integration of a conoscopic holography
sensor on a CMM. AIP Conf. Proc. 2012, 1431, 225–232. [CrossRef]
6. Enguita, J.M.; Álvarez, I.; Fraga, C.; Marina, J.; Fenández, Y.; Sirat, G. Conoscopic holography-based
long-standoff profilometer for surface inspection in adverse environment. Opt. Eng. 2006, 45, 073602.
[CrossRef]
7. Álvarez, I.; Enguita, J.M.; Frade, M.; Marina, J.; Ojea, G. On-line metrology with conoscopic holography:
Beyond triangulation. Sensors 2009, 9, 7021–7037. [CrossRef] [PubMed]
8. Post, D.; Han, B.; Ifju, P. High Sensitivity Moire: Experimental Analysis for Mechanics and Materials; Springer:
New York, NY, USA, 1994; ISBN 978-1-4612-4334-2.
9. Sciammarella, C.A.; Lamberti, L.; Boccaccio, A.; Sciammarella, F.M. High precision contouring with moiré
and related methods: A Review. Strain 2011, 47, 43–64. [CrossRef]
10. Chang, W.Y.; Hsu, F.H.; Chen, K.H.; Chen, J.H.; Hsieh, H.C.; Hsu, K.Y. Reconstruction of surface profile by
using heterodyne moiré method. Opt. Commun. 2012, 285, 5337–5340. [CrossRef]
11. Jeong, M.S.; Kim, S.W. Color grating projection moiré with time-integral fringe capturing for high-speed 3-D
imaging. Opt. Eng. 2002, 41, 1912–1917. [CrossRef]
12. Mu, J.; Guan, Z.; Kang, J.; Bian, T.; Su, F. Error analysis of phase shifting by varying the incident angle of
parallel beams in shadow moiré. Optik 2013, 124, 6769–6771. [CrossRef]
13. Sciammarella, C.A.; Lamberti, L.; Sciammarella, F.M. High-accuracy contouring using projection moiré.
Opt. Eng. 2005, 44, 093605. [CrossRef]
14. Nikon HN C3030. Available online: http://www.nikon.com/products/industrial-metrology/lineup/3d_
metrology/3d-coordinate-metrology/hnc3030/ (accessed on 5 December 2016).
15. Gleason GRSL—Gear Rolling Meets Advanced Laser Technology. Available online: https:
//www.gleason.com/en/products/metrology/metrology-systems/gear-rolling-systems/grsl-gear-
rolling-meets-advanced-laser-technology (accessed on 30 January 2019).
Sensors 2019, 19, 1450 16 of 16

16. Creath, K.; Schmit, J.; Wyant, C. Optical Metrology of Diffuse Surfaces. In Optical Shop Testing, 3rd ed.; Daniel
Malacara, Wiley-Interscience: Hoboken, NI, USA, 2007; pp. 756–779, ISBN 978-0-4714-8404-2.
17. Schreiber, H.; Bruning, J.H. Phase-Shifting Interferometry. In Optical Shop Testing, 3rd ed.; Daniel Malacara,
Wiley-Interscience: Hoboken, NI, USA, 2007; pp. 547–629, ISBN 978-0-4714-8404-2.
18. Hariharan, P.; Oreb, B.F.; Eiju, T. Digital phase-shifting interferometry a simple error-compensating phase
calculation algorithm. Appl. Opt. 1987, 26, 2504–2506. [CrossRef] [PubMed]

© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access
article distributed under the terms and conditions of the Creative Commons Attribution
(CC BY) license (http://creativecommons.org/licenses/by/4.0/).