© Queen’s Printer and Controller of HMSO, 2015.

3D Optical Scanner Dimensional Verification

Facility at the NPL’s “National FreeForm
Centre”
M. R. Dury, S. B. Brown, M. B. McCarthy, S. D. Woodward
National Physical Laboratory, UK

Email: martin.dury@npl.co.uk
Web: http://www.npl.co.uk/freeform

Abstract

Fringe projectors and other high precision 3D optical surface-form scanners are
rapidly being adopted by industry in favour of Cartesian CMMs because of their
advanced metrology benefits, which include point-cloud data capture, portability
and speed. However, certain surfaces and geometries are challenging for such
scanners to measure accurately, resulting in unreliable dimensional data. NPL
(National Physical Laboratory) has developed an environmentally controlled 3D
optical scanner dimensional verification facility to simulate typical usage
conditions where temperature and lighting may vary. In addition, a range of test
artefacts have been specifically developed to identify scanners’ sensitivity to
colour, resolution, roughness, and laser scanning articulating arm scan velocity.
This paper describes NPL’s verification facility that is to be launched within the
year and has been developed with the aim of providing a 3D optical scanner
verification service to global industry ensuring greater confidence in their
measurement capability.

1 Introduction

3D optical scanners, such as fringe projectors and laser scanning articulating

arms, are quickly taking the place of fixed tactile based CMMs as their
portability and rapid production of results make them particularly attractive
away from the metrology laboratory and onto the production line, as well as
outside at excavation sites and crime scenes. Having been developed in the
1950s [1], the performance of tactile CMMs is well understood [2], trusted and
established [3]. In contrast, 3D optical scanners are a relatively new and fast
developing technology, whose development has accelerated with increasing
© Queen’s Printer and Controller of HMSO, 2015.

access to low cost, powerful computers and cheaper optics. While 3D optical
scanners offer the potential of great positives, such as massively increased
measurement speeds, equipment portability and relative ease when measuring
freeform surfaces, their limitations are still being understood and international
standards that describe suitable tests and procedures for their acceptance and
use are yet to be developed, for example, VDI/VDE 2634 [4], the German
guideline for optical 3D measuring systems addresses neither freeform surfaces
nor surface finish.
The National FreeForm Centre, established in 2009, has developed a
purpose-built dimensional verification laboratory to assess 3D optical scanner
performance and environmental sensitivities with the aim of providing global
industry with greater confidence in their measurements. This paper reports the
development of NPL’s 3D optical scanner verification facility.

2 The 3D optical scanner verification facility

NPL’s 3D optical scanner verification facility comprises a purpose-built

environmentally-controlled laboratory (approximate dimensions: 3 m by 5 m
by 2.5 m), test artefacts, tests and test equipment that have been developed to
examine the performance of 3D optical scanners. With adjustable lighting
conditions and a temperature range from 16 °C to 24 °C, the facility is designed
to simulate the typical environments where 3D optical scanners are increasingly
being used. Fringe projectors commonly have ~(400×400×300) mm
measurement volumes with manufacturers claiming measurement accuracies
< 10 µm, while laser scanning articulating arms frequently work in ~33 m3
spherical volumes with claimed accuracies < 100 µm. To characterise such
instruments, NPL’s National FreeForm Centre have designed artefacts and test
procedures to obtain data quality (how representative the measured data is of
the object being measured) and data quantity (the amount of measured data)
information with regards to instrument environmental sensitivities, such as
temperature and lighting, instrument properties, such as resolution,
measurement volume and the effects of scanner orientation, and artefact
properties such as angle, colour, reflectance, roughness and material. Specific
laser scanning articulating arms tests also identify sensitivities to scan velocity,
height and joint encoder angle. The facility’s tests are described in more detail
below:

2.1.1 Temperature effects

Temperature is of critical importance when performing accurate dimensional

measurements, not only affecting the object being measured but also the
performance of 3D optical scanners [5]. Instrument manufacturers use different
methods to overcome temperature effects: some 3D optical scanners are
manufactured using carbon fibre, which is known for its low thermal expansion
coefficient [6], but may also exhibit hysteresis making its behaviour difficult to
predict [7]. Other manufacturers use aluminium as it has well-known thermal
© Queen’s Printer and Controller of HMSO, 2015.

properties and is therefore easier to predict than other materials.

Thermocouples embedded within an aluminium frame can be used to correct
for expansion or contraction due to temperature change and provide a warning
when the instrument’s calibration is likely to be affected.
3D optical scanner temperature tests at NPL are designed to measure
dimensional sensitivities to changing temperature. The tests comprise setting
the facility to a particular set point temperature, either 16 °C or 24 °C, and
placing the scanner inside it to stabilise. Once the scanner is at a stable
temperature, it is calibrated and set to continuously measure the ~300 mm
separation between two spheres attached to a Zerodur bar, shown in figure 1.
As dimensional measurements are taken, the temperature of the facility is
changed until it reaches the opposite set point (24 °C or 16 °C). To close the
measurement loop, the scanner is recalibrated and the measurements are
repeated while the facility’s temperature is returned to its original setting. The
dimensional data from the series of measurements is subsequently analysed to
identify changes in the measured separation distance between the sphere
centres.

Figure 1: Preliminary measurements showing the sphere artefact and the effect
of temperature on dimensional measurements from a fringe projector.

Figure 1 presents preliminary results from measurements on a fringe projector

which demonstrate the importance of understanding 3D optical scanner
temperature effects. The results, which will be reported in a future paper [8],
show a 6 µm per °C change in the separation between the two spheres.

2.1.2 Multi-faceted test artefact for normal and 45° measurements

The optical properties of the measurement object’s surface are known to affect
3D optical scanners [9]. Lambertian surfaces that diffusely reflect projected
light are ideal for 3D optical scanner measurements, while they may record
little useful data from specular surfaces that reflect projected light away like a
mirror. Surface colour may also affect the data that a 3D optical scanner is able
© Queen’s Printer and Controller of HMSO, 2015.

to record [10]. The data quantity may be affected as some 3D optical scanning
systems may have difficulty distinguishing a red laser line on a red surface or
dark fringes projected onto a dark surface. NPL has observed an embossing
effect between regions of high colour contrast when scanning flat surfaces that
have a multi-coloured pattern, suggesting that data quality is affected.
The National FreeForm Centre has developed multi-faceted test artefacts to
quantify 3D optical scanners’ ability to measure surfaces with different
reflectances, roughnesses and colours. In addition, surfaces of specific interest
to industries that include aerospace, automotive, manufacturing, medical and
heritage are addressed.

Figure 2: The NPL 2D material coupon plate.

Figure 2 presents one of the multi-faceted test artefacts, the NPL 2D material
coupon plate. This comprises eight artefact coupons surrounding a central,
highly Lambertian, sintered PTFE (Spectralon®) reference coupon. The 60 mm
by 60 mm square coupons include materials such as aluminium, carbon fibre
and titanium. The coupons are selected depending on the measurand, i.e. a
scanner’s ability to measure roughened surfaces, and are measured with the
calibrated scanner using optimal measurement parameters, at normal and ± 45°
orientations relative to the scanner’s projector. The point clouds from the
measurements are then analysed using software to identify a 30 mm by 30 mm
region within the centre of each coupon. The number of points within this
region is normalised against an identical region within the “white” central
Spectralon reference coupon to obtain a data quantity measurement and
therefore determine a scanner’s ability to measure a particular surface
reflectance, roughness, colour or material.

2.1.3 Multi-faceted test artefact for multiple angle measurements

Figure 3 presents NPL’s 3D material coupon plate that, like the NPL 2D
material coupon plate, is designed to assess a scanner’s ability to measure
© Queen’s Printer and Controller of HMSO, 2015.

different surfaces finishes, but at a greater number of angles with respect to the
scanner’s projector.

Figure 3: The NPL 3D material coupon plate.

NPL’s 3D material coupon plate comprises three rows and five columns of
coupons; a 150° angle separates each column rather than forming a plane. Five
Spectralon reference coupons are placed along the NPL 3D material coupon
plate’s central row, while the artefact coupons are placed in the rows above and
below. The NPL 3D material coupon plate is mounted on a rotation stage and
measured by the calibrated scanner with optimal measurement parameters,
through the range of angles of interest, with each column normal to the scanner
during one of the scans. As in the case of the NPL 2D material coupon plate
measurements, the point clouds from the NPL 3D material coupon plate
measurements are then analysed and the 30 mm by 30 mm regions within the
coupons and the Spectralon references are compared.

2.1.4 Illumination effects

3D optical scanners have been shown to be affected by the ambient lighting

[11, 12]. Changes in ambient lighting, flashes from welding processes and
projected shadows may all affect the measurements performed by 3D optical
scanners.

Figure 4: A fringe pattern projected onto the NPL 150 mm Freeform artefact.
© Queen’s Printer and Controller of HMSO, 2015.

NPL’s 3D optical scanner verification facility incorporates tests to quantify the

effects on measured dimensions (data quality) and data quantity of 3D optical
scanners in environments with different and changing illumination. In this
illumination test the scanner is initially calibrated under the conditions
recommended by the manufacturer and used to measure artefacts such as the
NPL 150 mm Freeform artefact [13] (shown in figure 4), the NPL 2D material
coupon plate and sphere artefacts without any ambient lighting. The artefacts
are then measured while a programmed set of different illuminations (lighting
levels, colours and patterns) are projected onto them. Each measurement point
cloud is then analysed to identify changes in artefact dimensions and data
quantity with changes in the illumination conditions.

2.1.5 3D Optical Scanner Measurement Resolution

As the take up of 3D optical scanners becomes more widespread in industry,

greater demands will be placed on them to measure ever finer detail. To help
3D optical scanner manufacturers verify their instruments, NPL have developed
the NPL Bessel plate artefact to provide a means of resolution measurement.
The Bessel plate artefact is a (170×170×45) mm aluminium plate with a
circular pattern cut into its top surface that is based upon the zero-order Bessel
function of the first kind, to provide a series of ripples with an exponential
decay in amplitude and also frequency with increasing distance from the plate’s
centre.

Figure 5: The NPL Bessel plate artefact.

3D optical scanner resolution is assessed by comparing scanner measurements

of the NPL Bessel plate with its CAD model, with particular attention paid to
the fine ripples on the outer regions of the plate.
© Queen’s Printer and Controller of HMSO, 2015.

2.1.6 Scanner Orientation

3D optical scanners are now being attached to robotic arms to create automated
measurement systems, particularly for the automotive industry [14-16]. The
scanners on these systems will be subject to a wide range of orientations and
slight movement of their optical components would change measurement
dimensions.
NPL’s National FreeForm Centre have developed a system to quantify
orientation effects on 3D optical scanners that uses a frame to ensure that a
sphere artefact retains its position relative to a calibrated scanner while the
scanner is oriented through a range of pitch and roll angles. The point clouds
from the measurements are analysed to identify changes in artefact dimensions
(sphere centre separation distances) with respect to scanner orientation.

2.1.7 Measurement Volume

Figure 6: The NPL tetrahedral and Bessel plate artefacts.

Lower cost 3D optical scanners are becoming increasingly available on the

market that, to be more affordable, may not incorporate the costly low
aberration optics of more expensive scanners. While the lower cost scanners
might be very capable of measuring artefacts placed at their measurement
volume’s centre, artefacts placed at its extremities may be more prone to
dimensional error. The National FreeForm Centre has developed a tetrahedral
artefact, shown in figure 6, to test for 3D optical scanner sensitivities to artefact
position within the measurement volume. The NPL tetrahedral is modular and
comprises coloured alignment spheres separated by changeable fixed-length
rods that provide sphere separations of 300 m, 600 m and 1000 m allowing it to
fit within a range of volumes; the coloured balls aid detection of colour
sensitivity effects possibly caused by chromatic aberration from the scanner’s
optics. While tetrahedral artefacts have already been developed to verify 3D
optical scanners [17], NPL’s tetrahedral is designed to hold the Bessel plate
© Queen’s Printer and Controller of HMSO, 2015.

artefact and allow simultaneous resolution, colour and measurement volume

tests.
To test a 3D optical scanner for dimensional changes within the
measurement volume, the tetrahedral is built to an appropriate size for the
instrument’s measurement volume and placed onto a rotation stage. The NPL
tetrahedral and Bessel plate are then measured by the scanner through a series
of rotations. The measurement point clouds are analysed to identify changes in
artefact dimensions (sphere diameter and sphere centres separation distances)
and any dimensional sensitivities to the colours of the spheres.

2.1.8 Laser Scanning Articulating Arm Scan Velocity, Height and

Joint Encoder Angle

Industrial measurements using laser scanning articulating arms may be a

compromise between scan velocity and measurement quality; slower movement
of the laser scanner may result in higher quality data and more dimensionally
accurate scans, but at a time cost. On the contrary, moving the scanner more
quickly over the measurement artefact may result in faster completion of the
scan, but at a cost to the measurement data quality and the dimensional
accuracy.

Figure 7: Preliminary measurement results showing increasing deviation in

measured sphere diameter and separation with scan velocity.

To find the optimum balance between scan velocity and measurement quality,
NPL’s National FreeForm Centre have developed a system where the
articulating arm’s scanning head is placed within an adjustable frame and
moved along by a precision carriage. To isolate the arm’s individual encoders
the frame can be moved relative to the scanner’s base and adjusted for height
and angle. A motor maintains a constant scanning head velocity during the
scan, thus removing operator effects such as varying velocity and height during
measurement. The effects of scan velocity, height and joint encoder angle on
data quality and data quantity are determined from the measurement point
© Queen’s Printer and Controller of HMSO, 2015.

clouds, by analysing the sphere dimensions, the separation between their

centres and the number of points that form them in the point cloud. Figure 7
presents preliminary results from the system that shows increasing deviation in
the measured sphere diameters and the separation of their centres with
increasing scan velocity, the results from this study will be published in a future
paper.

3 Summary

NPL’s National FreeForm Centre have developed a dimensional verification

facility to provide global industry with greater confidence in 3D optical
scanners, which are rapidly replacing traditional Cartesian CMMs as they offer
point-cloud data capture, portability and speed. When compared with CMMs,
3D optical scanners are a relatively new technology, their limitations are still
being understood and international standards regarding their use are still to be
developed. In addition, future ISO standards will involve the acceptance and
verification of 3D optical scanners, but will not address the range of issues that
NPL’s facility covers, such as the measurement of freeform surfaces.
NPL’s 3D optical scanner dimensional verification facility comprises a
purpose-built environmentally-controlled laboratory, artefacts, tests and test
equipment. The laboratory provides control over temperature and illumination
to assess environmental effects on scanner measurements. Artefacts have been
developed to assess scanner ability to measure the dimensions of objects with
different colours, reflectances, roughness, and materials of particular industrial
interest and to determine how these measurements are affected by angle. Other
artefacts have been designed to measure scanner resolution, their sensitivity to
orientation and their sensitivity to the object’s position within the measurement
volume. Tests to verify laser scanning articulating arm performance are being
developed to identify sensitivities to scan velocity, height and joint encoder
angle.
Work is ongoing within the facility which is to be launched within the year.
However, preliminary tests are already showing the effects on fringe projector
measurements with increasing temperature and laser scanning articulating arm
sensitivities to scan velocity. The results from both of these investigations and
other tests will be reported in future papers.

4 References

[1] COORD3. Coordinate Measuring Machine History – Fifty Years of

CMM History leading up to a Measuring Revolution.
http://www.coord3-cmm.com/50-years-of-coordinate-measuring-
machine-industry-developments-and-history/ (accessed 3 December
2014).
[2] Flack D, 2011 CMM verification (National Physical Laboratory) Good
Practice Guide No 42 http://www.npl.co.uk/publications/cmm-
verification.
© Queen’s Printer and Controller of HMSO, 2015.

[3] ISO 10360-2:2009 Geometrical product specifications (GPS) --

Acceptance and reverification tests for coordinate measuring
machines (CMM) -- Part 2: CMMs used for measuring linear
dimensions
[4] VDI 2634 Optical 3D-measuring systems Berlin Beuth 2008
[5] Adamczyk M, Kaminski M, Sitnik R, Bogdan A, Karaszewski M.
2014 Appl Optics.53 5154-62.
[6] Rogers KF, Kingston-Lee DM, Phillips LN, Yates B, Chandra M,
Parker SFH. 1981 J Mater Sci.16 2803-18.
[7] Tompkins SS. 1987 Int J Thermophys.8 119-32.
[8] Dury M, Brown S, McCarthy M, Woodward S. Blowing hot and cold:
temperature sensitivities of 3D optical scanners. Abstract submitted to
euspen's 15th International Conference & Exhibition, Leuven, June
2015.
[9] Levoy M, et al. The digital Michelangelo project: 3D scanning of
large statues Siggraph 2000 (New Orleans, USA, 23 - 28 July 2000)
(SIGGRAPH 2000 Conference Proceedings) (New York: Assoc
Computing Machinery) pp. 131-44
[10] Clark J, Robson S. 2004 Surv Rev.37 626-38.
[11] Voisin S, Foufou S, Truchetet F, Page D, Abidi M. 2007 Opt Eng.46.
[12] Lemeš S, Zaimović-Uzunović N. Study of ambient light influence on
laser 3D scanning. 7th International Conference on Industrial Tools
and Material Processing Technologies; 4 - 7 October 2009; Ljubljana,
Slovenia. pp. 327-30
[13] McCarthy M, Brown S, Evenden A, Robinson A 2011 NPL freeform
artefact for verification of non-contact measuring systems Three-
Dimensional Imaging, Interaction, and Measurement (San Francisco,
USA, 24 - 27 January 2011) (Proceedings of SPIE vol 7864) ed
Beraldin J, et al.
[14] AICON. Employing robots to shorten development cycles.
http://aicon3d.com/latest-news/news/news-in-detail/article/employing-
robots-to-shorten-development-cycles.html (accessed 3 December
2014).
[15] GOM. ATOS ScanBox - Optical 3D Measuring Machine.
http://www.gom.com/metrology-systems/system-overview/atos-
scanbox.html (accessed 3 December 2014).
[16] Steinbichler. Metrologic Group Automates the STEINBICHLER T-
SCAN CS Laser Scanner.
http://blog.steinbichler.com/2014/05/02/metrologic-group-automates-
the-steinbichler-t-scan-cs-laser-scanner/ (accessed 3 December 2014).
[17] Acko B, McCarthy M, Haertig F, Buchmeister B. 2012 Meas Sci
Technol.23.