JANUARY 2018

www.vision-systems.com

VISION AND AUTOMATION SOLUTIONS

FOR ENGINEERS AND INTEGRATORS WORLDWIDE

LED strobe
controllers
When should you invest

Embedded
camera
Brings 3D mapping to
tunnel boring machines

High-speed
cameras
Target machine vision
applications

Deep
learning
Enhances
machine
vision

1801vsd_C1 1 1/4/18 8:07 AM

1801vsd_C2 2 1/4/18 8:07 AM
–

January 2018
VOL. 23 | NO. 1
Vision and Automation Solutions for
Engineers and Integrators Worldwide A P EN N W EL L P U B L I C AT I O N

C ove r St o r y
A universal, pre-trained classifier based on deep
learning algorithms enables identification of a
wide range of OCR fonts. (See page 15).

f e ature s
15 IN T EGR AT ION INSIGH T S d ep ar tment s
How deep learning is enhancing machine vision
3 My View
Developers increasingly apply deep learning and
4 Online@vision-systems.com
artificial neural networks to improve object detection
5 Snapshots
and classification. Read the latest news from our website
Johannes Hiltner 9 Technology Trends
ELECTRONICS INSPECTION
18 IN T EGR AT ION INSIGH T S Infrared imaging microscopes aid in
detecting integrated-circuit defects
When to invest in LED strobe controllers DEEP-SEA IMAGING

LED strobe controllers offer many benefits for Imaging system monitors for pollution
and coral health in the Bay of Biscay
applications involving low contrast, motion, mark 3D IMAGING
quality assessment, OCR and ambient lighting issues. 3D imaging enables high-speed
precision bore hole inspection
Alan L. Lockard
29 Vision+Automation Products

32 Ad Index/Sales Offices
21 INDUS T RY S OLU T IONS PROFIL E

Embedded camera brings 3D mapping

to tunnel boring machines
Mounting an embedded camera within the cutting
head of a tunnel boring machine allows contractors to
document rock face structures as boring progresses.

Robert Wenighofer

25 PRODUC T F OC US • Complete Archives • White Papers

High-speed cameras target machine • Industry News • Feedback Forum

vision applications • Buyers Guide • Free e-newsletter
Faced with imaging high-speed events, developers
• Webcasts • Video Library
can choose from those that transfer data over
industry-standard camera-to-computer interfaces www.vision-systems.com
and stand-alone cameras.

Andrew Wilson

www.vision-s y stems .com VISION SYSTEMS DESIGN Januar y 2018 1

1801vsd_1 1 1/4/18 7:59 AM

 Visit us at

Jan. 27 - Feb. 1, 2018 | Booth 823

1801vsd_2 2 1/4/18 7:59 AM

 my
view
Alan Bergstein: Group Publisher
(603) 891-9447
alanb@pennwell.com
John Lewis: Editor in Chief
(603) 891-9130
johnml@pennwell.com
James Carroll Jr.: Senior Web Editor
(603) 891-9320
jamesc@pennwell.com
Andrew Wilson: European Editor
+44 7462 476477
What is deep learning?
andycharleswilson@gmail.com
Robert Tait: Contributing Editor While sometimes seemingly used interchangeably, the subtle differences in
518-269-9410
tait@opticalmetrologysolutions.com meansings between artificial intelligence (AI), machine learning and deep
Kelli Mylchreest: Art Director learning may cause some confusion. In contrast to natural intelligence dis-
Mari Rodriguez: Production Director played by humans and other animals, AI refers to machines mimicking human
Dan Rodd: Senior Illustrator
Debbie Bouley: Audience Development Manager
cognitive functions such as problem solving or learning. So when a machine
Alison Boyer Murray: Ad Services Manager understands human speech or can compete with humans in a game of chess, AI applies.
Joni Montemagno: Marketing Manager Machine learning is the current state-of-the-art application of AI and is largely responsible
for its recent rapid growth. Based on the idea of giving machines access to data so that they can
learn for themselves, machine learning has been enabled by the internet, and the associated rise
www.pennwell.com in digital information being generated, stored and made available for analysis. Building on AI
EDITORIAL OFFICES: concepts, machine learning focuses on solving real-world problems through architectures such
Vision Systems Design
61 Spit Brook Road, Suite 401 as artificial neural networks designed to imitate human decision making.
Nashua, NH 03060 Deep learning concentrates on a subset of machine-learning techniques, with the term “deep”
Tel: (603) 891-0123 Fax: (603) 891-9328
www.vision-systems.com generally referring to the number of hidden layers in the deep neural network. While a conven-
CORPORATE OFFICERS: tional neural network may contain a few hidden layers, a deep network may have tens or hun-
Robert F. Biolchini, 1939-2017: Chairman dreds of layers. In deep learning, a computer model learns to perform classification tasks directly
Frank T. Lauinger: Vice Chairman
from text, sound or image data. In the case of images, deep learning requires substantial com-
Mark C. Wilmoth: President and
Chief Executive Officer puting power and involves feeding large amounts of labeled data through a multi-layer neural
Jayne A. Gilsinger: Executive Vice President,
Corporate Development and Strategy
network architecture to create a model that can classify the objects contained within the image.
Brian Conway: Senior Vice President, Major technology companies such as Google, Facebook, IBM, Intel and Microsoft have
Finance and Chief Financial Officer
invested in deep learning for some time, but more recently, machine-vision software compa-
TECHNOLOGY GROUP:
Christine A. Shaw: Senior Vice President and nies have begun to apply deep learning within their products, or base their entire product on it.
Group Publishing Director Machine vision has predominantly relied on rules-based machine-vision algorithms that excel
FOR SUBSCRIPTION INQUIRIES in applications where you know exactly what you’re looking for. Classic edge detection, blob,
Tel: (800) 869-6882
Fax: (866) 658-6156
object- and feature-location algorithms generally excel in tasks requiring sub-pixel accuracy for
email: : VSD@kmpsgroup.com precision measurements or robot guidance.
web: www.vsd-subscribe.com
The value of deep learning in machine-vision applications stems from its ability to make
Vision Systems Design® (ISSN 1089-3709), Volume human-like judgments of part quality and other example-based decisions. Verifying the presense
23, No.1 Vision Systems Design is published 11 times
a year in January, February, March, April, May, June, of bolts, brackets, foam pads and straps on car seat assemblies, for example, can challenge tradi-
July/August, September, October, November, December
by PennWell® Corporation, 1421 S. Sheridan, Tulsa, OK tional machine vision systems if subcomponents come from a variety of suppliers with variations
74112. Periodicals postage paid at Tulsa, OK 74112 and
at additional mailing offices. SUBSCRIPTION PRICES: in color and texture. In such applications, deep learning helps machine-vision systems cope with
USA $130 1yr., $190 2 yr., $244 3 yr.; Canada $148
1 yr., $217 2 yr., $280 3 yr.; International $160 1 yr.,
the range of acceptable part appearances. Likewise, our cover story this month on page 15 dis-
$235 2 yr., $305 3 yr. POSTMASTER: Send address cusses how a single, universal, pre-trained classifier based on deep-learning algorithms enables
corrections to Vision Systems Design, P.O. Box 47570,
Plymouth, MN 55447. Vision Systems Design is a reg- identification of a wide range of typefaces in optical character recognition applications. Also, on
istered trademark. © PennWell Corporation 2017. All
rights reserved. Reproduction in whole or in part with- page 7 we cover development of a neural network designed to identify six different defect classes
out permission is prohibited. Permission, however, is
granted for employees of corporations licensed un- on reflective metallic surface images. As always, I hope you enjoy this issue.
der the Annual Authorization Service offered by the
Copyright Clearance Center Inc. (CCC), 222 Rosewood
Drive, Danvers, Mass. 01923, or by calling CCC’s
Customer Relations Department at 978-750-8400 pri-
or to copying. We make portions of our subscriber list
available to carefully screened companies that offer
products and services that may be important for your
work. If you do not want to receive those offers and/
or information via direct mail, please let us know by John Lewis, EDITOR IN CHIEF
CH
contacting us at List Services Vision Systems Design, 61
WWW.VISION-SYSTEMS.COM
WWW VISION SYSTEMS COM
Spit Brook Road, Suite 401, Nashua, NH 03060. Printed
in the USA. GST No. 126813153. Publications Mail
Agreement no. 1421727.

www.vision-s y stems .com VISION SYSTEMS DESIGN Januar y 2018 3

1801vsd_3 3 1/4/18 7:59 AM

–

online @ www.vision-systems.com
3D Time of Flight company 10 ways deep learning Machine vision standard
Odos Imaging acquired by technology is being directs on evaluation of
Rockwell Automation deployed today image processing systems
Odos Imaging, a de- As the demands for speed A new standard puts
veloper of 3D Time of and accuracy in machine forward procedures for
Flight cameras for indus- vision and image process- assessing classificatory
trial imaging applications, ing applications grow, performance during the
has been acquired by Rockwell Automa- so too must the software that is used in acceptance of machine vision systems,
tion, Inc. http://bit.ly/VSD-1801-01 them. http://bit.ly/VSD-1801-03 while also providing examples from in-
dustrial inspection technology, which
Euresys acquires OEM Lucid Vision Labs machine serves as guidance.
imaging IP and hardware vision camera company http://bit.ly/VSD-1801-05
company Sensor to Image founded by industry veteran
Euresys has announced Lucid Vision Labs, Inc., High-speed imaging sheds
the acquisition of Sensor a company that designs light on secrets of flight
to image GmbH, a com- and manufactures ma- Using high-speed imag-
pany specializing in OEM chine vision cameras and ing and computation-
imaging IP and hardware for industrial components, has been founded by former al fluid dynamics, the
vision products, including cameras and Founder, President and VP of Engineering unique mechanisms in-
frame grabbers. at Point Grey Research Inc. Rod Barman. volved in mosquito flight have been re-
http://bit.ly/VSD-1801-02 http://bit.ly/VSD-1801-04 vealed. http://bit.ly/VSD-1801-06

tdtsefifeeffieptleeseeeetseefeeeee
sl pulsed ptwer laser
pselatastspecaalazedtantdevelppangptcustptazangtt
andttanuuacturangtlaser patterr gererattrstt
uprttachanetvasapntapplacatapnss
Thanks for a great 2017!
We wish you a Happy New Year.

putputtppwer
tr tr tr tr
Discover more:
Thank you!
www.alysium.com
tFF tFF tFF
eeeteeeteeeeeeeeteeeeeeteeefieffifieeee

ueatures
pfeeeteepeetpppptppptfieptpppeetfeteetpw seeeeetefitefeeeeeee
aeeeefifieeptteefieeseetefeeeteffifipteefit seeeeefieeeteefieeetfieptpeefieeeteeeetseest
eeeeepteeaeetefieeeeeeee pfeyteyeeetfeteetpp%

structuredtlaghttandtt
lasertbeattshapangtsplutapns
ppp9tt2eptaaeefe
lfieseeeptlcpthpttttpptcfiefipfi

eeee@eeeefiseeett ttppptptps2227tt
osela.com

4 Januar y 2018 VISION SYSTEMS DESIGN www.vision-s y stems .com

1801vsd_4 4 1/4/18 8:00 AM

 snapshots Short takes on the
leading edge

Fully self-driving
cars are here,
says Waymo
Waymo’s (Mountain View, CA, USA; www.waymo.com) fully self-
driving vehicles are now test driving in fully autonomous mode on
public roads without a driver after having worked on the technology
for more than eight years.
A subset of vehicles from Waymo—formerly the Google self-driv-
ing car project—will operate in fully autonomous mode in the Phoe-
nix metro region. Eventually, the fleet will cover a region larger than
the size of Greater London, with more vehicles being added over time.
Waymo explains in a press release that it has prepared for this next tests, practicing rare and unusual cases,” wrote Waymo via Medium.
phase by putting its vehicles through the world’s longest and tough- “We’ve multiplied all this real world experience in simulation, where
est ongoing driving test. our software drives more than 10 million miles every day. In short: we’re
“Since we began as a Google project in 2009, we’ve driven more than building our vehicles to be the most experienced driver on the road.”
3.5 million autonomous miles on public roads across 20 U.S. cities. At Waymo’s vehicles, according to the company, are equipped with
our private test track, we’ve run more than 20,000 individual scenario the necessary features, including sensors and continued on page 6

Standalone vision system targets small parts inspection

Machine vision solution provider and sys-
tems integrator Artemis Vision (Denver, CO,
USA; www.artemisvision.com) has developed
the visionStation standalone vision system,
which is designed to perform the inspection
of small parts.
When it comes to small parts
inspection, there are number of
challenges that must be over-
come, starting with space con-
straints. Some small parts
manufacturers have
limited space for in-
spection equipment,
so it is important to understand
the manufacturer’s needs
and space require-
ments. Accuracy and
repeatability must also be considered. In-
specting small parts by eyesight alone can be
a daunting task, and maintaining a consis-
tent method or process of inspection can
change from person to person. Small
parts inspection applications must be
consistent for them to be success-
fully accurate and repeatable.
Lastly, for small parts inspection
to be successful, the
manufacturer needs
to know what defects
they are looking for, as
some defects can be so
small that they go un-
noticed. Additionally,
manufacturers need to
catalog known defects,
which becomes easy when machine vision is
used, as opposed to human inspection.
Artemis Vision’s visionStation is precision
machined from aluminum and fitted with
a digital machine vision camera and light-
ing that is tailored to suit the type of inspec-
tion performed. The system is an operated-
loaded unit that can be used as a standalone
inspection station, or integrated into an exist-
ing operation.
“The visionStation is ideal for customers
who need to run approximately 20,000 parts
per day or less,” stated Tom Brennan, Pres-
ident at Artemis Vision. “At 3 seconds per
cycle, an operator can load about 10,000 parts
per shift. Customers who need flexible auto-
mation and repeatability between operators
and production batches continued on page 7

www.vision-s y stems .com Januar y 2018 5

1801vsd_5 5 1/4/18 8:00 AM

 snapshots
continued from page 5
software, to provide full autonomy, includ-
ing backup steering, braking, computer, and
power that can bring the vehicle to a safe stop,
if needed. The vehicles’ sensor suite includes
the following:
LiDAR (Light detection and ranging): This
sensor beams out millions of laser pulses per
second in 360° to measure how long it takes to
reflect off a surface and return to the vehicle.
Waymo’s system includes three types of
LiDAR developed in-house: a short-range
LiDAR that provides an uninterrupted view
directly around it, a high-resolution mid-
range LiDAR, and a next generation long-
range LiDAR that can see almost three foot-
ball fields away.
Vision systems: The vision system includes
360° field of view cameras that detect color
in order to spot things like traffic lights, con-
struction zones, school buses, and the flash-
ing lights of emergency vehicles. The system
is comprised of several sets of high-resolution
cameras that operate at long range in daylight
and low-light conditions.
In addition to LiDAR and a vision system,
the vehicles feature a radar system with a con-
tinuous 360° view and supplemental sensors
including audio detection and GPS.
“With Waymo in the driver’s seat,” accord-
ing to a statement, “we can reimagine many
different types of transportation, from ride-
hailing and logistics, to public transport and
personal vehicles, too.”
“We’ve been exploring each of these areas,
with a focus on shared mobility. By giving
people access to a fleet of vehicles, rather than
starting with a personal ownership model,
more people will be able to experience this
technology, sooner. A fully self-driving fleet
can offer new and improved forms of sharing:
it’ll be safer, more accessible, more flexible,
and you can use your time and space in the
vehicle doing what you want.”
The first application of Waymo’s fully-self
driving technology will be a Waymo driver-
less service. Over the next few months, Waymo
will be inviting members of the public to take
trips in its fully self-driving vehicles.
View a video of the driverless car in action
here: http://bit.ly/VSD-WAYMO
6 Januar y 2018
1801vsd_6 6
New specification from MIPI Alliance
streamlines integration of image
sensors in mobile devices
The MIPI Alliance (www.mipi.org)—a
global organization that develops inter-
face specifications for mobile and mobile-
influenced industries—has released a new
specification that provides a
standardized way to integrate
image sensors in mobile-con-
nected devices.
Named MIPI Camera Com-
mand Set v1.0 (MIPI CCS v1.0),
the specification defines a stan-
dard set of functionalities for
implementing and controlling
image sensors. This specifica-
tion is offered for use with MIPI
Camera Serial Interface 2 v2.0 (MIPI CSI-2
v2.0) and is now available for download. Ad-
ditionally, in an effort to help standardize use
of MIPI CSI-2, MIPI Alliance membership is
not required to access the specification.
MIPI CSI-2, according to the alliance, is
the industry’s most widely-used hardware in-
terface for deploying camera and imaging
components in mobile devices, including
drones. The introduction of MIPI CCS to
MIPI CSI-2 provides further interoperabil-
ity and reduces integration time and costs
for complex imaging and vision systems, ac-
cording to the alliance.
“MIPI Alliance is building on its success
in the mobile camera and imaging ecosys-
tem with MIPI CCS, a new specification
that will enhance the market-enabling con-
veniences MIPI CSI-2 already provides,”
said Joel Huloux, chairman of MIPI Alli-
ance. “The availability of MIPI CCS will
help image sensor vendors promote greater
adoption of their technologies and it will
help developers accelerate time-to-mar-
ket with innovative designs targeting the
mobile industry, connected cars, the Inter-
net of Things, AR/VR and other areas.”
With imaging applications becoming
more sophisticated and manufacturers de-
VISION SYSTEMS DESIGN
ploying multiple image sensors in their
products, implementation becomes more
complex and time consuming. MIPI CCS
aims to address these issues by making it
possible to craft a common software driver
to configure the basic functionalities of any
off-the-shelf image sensor that is compliant
with MIPI CCS and MIPI CSI-2 v2.0, ac-
cording to the MIPI Alliance. The specifica-
tion provides a complete command set that
can be used to integrate basic image sensor
features, including resolution, frame rate
and exposure time, as well as advanced fea-
tures like phase detection auto focus, single
frame HDR or fast bracketing.
“The overall advantage of MIPI CCS is
that it will enable rapid integration of basic
camera functionalities in plug-and-play fash-
ion without requiring any device-specific
drivers, which has been a significant pain
point for developers,” said Mikko Muukki,
technical lead for MIPI CCS. “MIPI CCS
will also give developers flexibility to cus-
tomize their implementations for more ad-
vanced camera and imaging systems.”
The new specification was developed for
use with MIPI CSI-2 v2.0 and is backward
compatible with earlier versions of the MIPI
CSI-2 interface. It is implemented on either
of two physical layers from MIPI Alliance:
MIPI C-PHY or MIPI D-PHY.
View more information on the specifica-
tion: http://bit.ly/VSD-MIPI
www.vision-s y stems .com
1/4/18 8:00 AM

continued from page 5
without necessarily wanting or needing
full inline automation where parts are
conveyed on a belt.
Within the visionStation is a Prosil-
ica GT4905 machine vision camera from
Allied Vision (Stadtroda, Germany; www.
alliedvision.com). This camera features the
KAI-16050 CCD image sensor from ON
Semiconductor (Phoenix, AZ, USA; www.
onsemi.com), which is a 16 MPixel sensor
that achieves a frame rate of 7.5 fps through
GigE interface. This camera was selected,
according to Brennan, as a result of its “high
resolution and excellent dynamic range.”
“The dynamic range plays an impor-
tant part when imaging the shiny areas
of a part, created by the lighting source.
Furthermore, the Prosilica GT4905 can
handle the rigorous conditions the vision-
Station may be used in,” he said.
Software for the system includes Omron
Microscan’s (Renton, WA, USA; www.mi-
croscan.com) Visionscape and Artemis Vi-
sion’s proprietary visionWrangler software.
“Our software is built from our exten-
sive library of components – reading a data
matrix, doing measurements, absence-
presence and counts,” explained Brennan.
The visionStation is offered in two
sizes, the small and larger system measur-
ing 1 and 1.5 cubic ft, respectively. The
small system can accommodate parts up
to 2 in. (5.08 cm), while the large system
can hold parts up to 6 in. (15.24 cm). The
minimum defect size allowance for each
system size is: 0.003 in. (76.2 µm) for the
small station and 0.01 inches (254 µm) for
the larger station.
Artemis Vision builds visionStation sys-
tems based on the parts a manufacturer
supplies and intends to have inspected.
In the next generation visionStation,
Artemis Vision reportedly plans to up-
grade its sensor.
“We will likely upgrade to a newer
CMOS camera and use C-mount optics,”
Brennan commented, “which will pro-
vide more lensing options and reduce the
overall size of the unit.”
www.vision-s y stems .com VISION SYSTEMS DESIGN
1801vsd_7 7
Deep learning and convolutional
neural networks
Deep learning with neural networks provide
significant image processing benefits regard-
ing object classification, image analysis and
image quality. Because small neural networks
suffice for many typical machine vision appli-
cations, processors such as FPGAs (field pro-
grammable gate arrays) can be implemented
effectively for convolutional neural networks
(CNNs), resulting in application expansion
beyond current classification tasks and effi-
cient use within embedded vision systems.
Deformed test objects, irregular shapes,
object variations, irregu-
lar lighting and lens dis-
tortions push the limits
of classic image process-
ing when framework
conditions for image ac-
quisition cannot be con-
trolled. Even individual
algorithms for feature
description are often
barely possible.
CNNs, on the other
hand, define characteris-
tics through its training
method, without using
mathematical models,
which make it possible
to capture and analyze Rolled-in Patches
scale
images in difficult situa-
tions such as reflecting surfaces, moving ob-
jects, face detection and robotics, especially
when easier classification of image data di-
rectly from the preprocessing to the classifica-
tion result is required.
Nevertheless, CNNs cannot cover all areas
of classic image processing, such as precise lo-
cation determination of objects. Here, new
and advanced CNNs must be developed. Prac-
tical experience with CNNs from prior years
led to mathematical assumptions and simpli-
fications (pooling, ReLu and overfitting avoid-
ance, to name a few) which led to a reduction
in computational expense, which in turn en-
abled implementation of deeper networks.
By reducing image depth at the same rate
of detection and by optimizing the algorithm,
snapshots
CNNs can be significantly accelerated and are
now ideally suited for image processing. Be-
cause CNNs are shift and partially scale in-
variant, they allow use of the same network
structures for different image resolutions, and
smaller neural networks are often sufficient for
many image processing tasks.
Due to the high degree of parallel pro-
cessing, neural networks are particularly well
suited to FPGAs, upon which CNNs also an-
alyze and classify high-resolution image data
in real time. In machine vision, FPGAs func-
Crazing Pitted Inclusion Scratches
surface
tion as massive accelerators of image process-
ing tasks and guarantee real-time processes
with deterministic latencies.
Until now, the high programming effort
and the relatively-low resources available in
an FPGA hindered efficient use. Algorithmic
simplifications now make it possible to con-
struct efficient networks with high throughput
rates in an FPGA.
To implement CNNs on FPGA hardware
platforms, the VisualApplets graphical environ-
ment from Silicon Software (Mannheim, Ger-
many; www.silicon.software) can be used. The
CNN operators in VisualApplets allow users to
create and synthesize diverse FPGA application
designs without hardware programming experi-
ence in a short time. continued on page 8
Januar y 2018 7
1/4/18 8:00 AM
 snapshots

continued from page 7 By transferring the weight and gradient param-

eters determined in the training process to the CNN operators, the
FPGA design is configured for the application-specific task. Opera-
tors can be combined into a VisualApplets flow diagram design with
digital camera sources as image input, and additional operators to op-
timize image preprocessing.
For larger neural networks and more complex CNN applications,
a new programmable frame grabber in the microEnable marathon
series is being released that has a 2.5-times more FPGA resources com-
pared to the current marathon series. The new frame grabber is ex-
pected to be suitable for neural networks with more than 1GB/sec
CNN bandwidth.
The CNNs are not only capable of running on the frame grabbers’
FPGAs but also on VisualApplets compatible cameras and vision sen-
sors. Since FPGAs are up to ten times more energy-efficient compared
to GPUs, CNN-based applications can be implemented particularly
well on embedded machine vision systems or mobile robots with the
required low heat output.
Application diversity and complexity with neural networks based on
newly-developed processors is expected to increase. A cooperation and
exchange of research results with Professor Michael Heizmann from the
Institute for Industrial Information Technology (IIIT) at the Karlsruhe
Institute of Technology (KIT; Karlsruhe, Germany; https://www.kit.edu/
english), part of the FPGA-based “Machine Learning for Industrial Ap-
plications” project, will generate future hardware and software advances.
To determine the percentage of correctly-detected defects in diffi-
cult environmental conditions, a neural network with 1,800 reflective
metallic surface images was trained on which of six different defect
Ruggedized Mechanical Design Minimizes  classes were defined. Large differences in scratches coupled with small
Image Aberrations from Shock and Vibration differences in crazing, paired with different surface grey tones from
lighting and material alterations, made the analysis of the surface
almost impossible for conventional image processing systems.
Redundant  All elements  Same optical  The test results demonstrated that the different defects were posi-
Ultra design as 
focus locks:  are bonded  Compact 
lock nut plus  inside of the  Computar's 
locking screw lens housing
Design MPW2 Series  tively classified by the neural network an average of 97.4%, which is
(2/3” 5MP)
a higher value compared to classic methods. The data throughput in
this application configuration was 400MB/sec. By comparison, a CPU-
based software solution achieved an average 20MB/sec.
The implementation of deep learning on FPGA processor technol-
ogy, which is ideally suited for image processing, is an important step.
Requirements that the method must be deterministic and algorithmi-
cally verifiable, however, will make entry into all areas of image pro-
cessing more difficult. Likewise, options to document which areas were
identified as errors, as well as segmentation and storage, have not yet
been implemented.
Thus far, training and operational use of CNNs are two separate pro-
cesses. In the future, new generations of FPGAs with higher resources
computar.com or using powerful ARM / CPU and GPU cores will enable on-the-fly
training on the newly acquired image material, which will further
EAST COAST: WEST COAST: MEXICO: increase the detection rate and will simplify the learning process.
1 (919) 230-8700 +1 (310) 222-8600 +52 (55) 5280 4660 This article was written by Martin Cassel, Silicon Software
(Mannheim, Germany; www.silicon.software)

8 Januar y 2018 VISION SYSTEMS DESIGN www.vision-s y stems .com

1801vsd_8 8 1/4/18 8:00 AM

 technologytrends
John Lewis, Editor, johnml@pennwell.com

• DEEP-SEA I M AG I N G
ELECTRONICS INSPECTION

Imaging systems monitor for Infrared imaging

pollution and coral health
Scientists from the Ifremer (French Research
Institute for the Exploitation of the Sea; Issy-
les-Moulineaux, France; wwz.ifremer.fr/en)
have used FCB block cameras from Sony Corp.
(Tokyo, Japan; http://pro.sony.com) to monitor
for pollution and coral health in the vulnerable
marine ecosystems of the Bay of Biscay.
After surveying 24 of the Bay’s canyons, and
three locations between adjacent canyons, the
images enabled identification of 11 coral hab-
itats, containing 62 coral species, at depths
between 50 and 1,000 meters. Surveying 15
canyons and three sites on the edge of the con-
tinental shelf, the team analyzed 6255 images
from the camera and recorded 198 items of
unique litter.
According to the team’s published findings,
litter appeared to accumulate at water depths
of 801 - 1100 m and 1401 - 1700 m and was
found in all canyons and continental shelfs
surveyed. More importantly, the research-
ers pointed out that 15 to continued on page 12
microscopes aid in
detecting integrated
circuit defects
Radiant Optronics (Singapore, Singapore;
www.radiantoptronics.com) has devel-
oped an infrared (IR) microscope that is
used to inspect integrated circuits (IC) for
internal defects or cracks that may occur
during the manufacturing process.

• 3D I M AG I N G

3D imaging enables high-speed

bore inspection
New inline-inspection applications are being developed for industry thanks to advances in
high-speed, low-coherence interferometry, a high-resolution three-dimensional (3D) imag-
ing technology. Precision bore holes, for example, must adhere to strict geometric dimen-
sioning and tolerancing (GD&T), including straightness, cylindricity, circularity, taper,
distortion, runout, roughness characteristics, semi-transparent coating thickness, or defect Radiant Optronics has developed an IR
microscope fitted with a SWIR camera that
characterization. Also, features on the insides of bores, such as threads, O-ring grooves,
inspects ICs for internal defects or cracks.
undercuts, chambers, and cross-holes, must often be characterized with high precision to
ensure that components ultimately interlock, seal, fit, or otherwise function as intended. Wafer is a thin substrate of semicon-
Understanding this, Novacam Technolo- ductor material that is used in electronics
gies (Pointe-Claire, QC, Canada; www.nova- for IC fabrication. There are many types
cam.com) has developed a system, dubbed of semiconductor materials and one of
BoreInspect, that provides fast, micron-pre- the most commonly materials used in
cision, non-contact measurements of bore electronics is Silicon (Si).
internal diameter (ID) surfaces. Silicon wafer, a key component in ICs,
is formed from highly pure, nearly defect-
A 3D imaging system based on high-speed
low-coherence interferometry is being used free, monocrystalline silicon sliced from a
for the automated inspection of bore holes. silicon boule. Serving as the substrate for
continued
continuedon
onpage
page10
9 microelectronic devices continued
continuedononpage
page139

www.vision-s y stems .com VISION SYSTEMS DESIGN Januar y 2018 9

1801vsd_9 9 1/4/18 8:00 AM

 technologytrends
The system automatically generates and presents the operator with a display of the four
bores, along with an evaluation of crucial measurement criteria.
continued from page 9 Vuk Bartulovic, Pres-
ident of Novacam, says “By the time they
come to us, many of our clients have spent
a long time looking for a way to measure, in
a non-contact manner, internal diameters
of bores, cylinders, and other such hard-to-
reach spaces. These bore diameters range
from a few millimeters to as large as truck
engine cylinder bores. Typically, they want
to acquire full internal diameter dimensions,
detect defects, and measure roughness. It is a
revelation to them that an instrument exists
that can provide all this functionality … and
that micron-precision data on their bore ID
can be acquired in a matter of seconds and in
a non-contact manner.”
The system has been deployed for auto-
mated 3D metrology of valve-body assembly
bores. Following a user-created inspection
sequence, all four bores in the valve body
are automatically scanned, and the resultant
3D surface point-cloud analyzed and dis-
played using PolyWorks from InnovMetric
(Québec, QC, Canada; www.innovmetric.
com), a third-party 3D data visualization and
metrology software. According to user crite-
ria, selected aspects of each bore are com-
pared to actual measurement data and non-
compliance visually highlighted.
Comprised of three main hardware com-
ponents, BoreInspect includes a rotational
scanner that spins a non-contact optical probe
while a robot or motion stage advances it into
the bore hole. The side-facing probe directs a
low-coherence light beam at the surface and
collects reflected light signals. The second
main component, a profilometer, provides
the light source to the rotational scanner via
a fiber-optic cable and processes the optical
signal received from the scanner. Finally, a
10 Januar y 2018
1801vsd_10 10
PC provides the user interface for data acqui-
sition control and data-analysis software. It
also typically houses client-selected 3D- visu-
alization and metrology software.
High-density point clouds generated from
its scans enable rapid and detailed charac-
terization of inner surfaces and features as
well as defect detection. During operation,
the scanner spins the probe at 1,800 rpm,
acquiring 30,000 3D surface measurements
per second. In just over 3 seconds, 100,000
surface measurements on the inside of the
bore are collected, providing a high-fidelity
3D model of the bore ID.
For the automated valve body bore
metrology application, the rotational scan-
ner enters each one of the four valve body
bores in an automated sequence and the
system automatically generates and pres-
a)
Defects
Cross-holes
Bore hole circumference (15.5 mm)
b)
Defect
Cross-holes
Along with a 3D point cloud, 3D height data
(top), and a light intensity image (bottom), can
be obtained and used for defect detection.
VISION SYSTEMS DESIGN
To help manufacturers gain insight into the
process, BoreInspect can be configured to
provide deviation maps highlighting the devi-
ation of the measured bore dimensions from
the CAD model, to reveal micron-level sur-
face defects such as this 19.7-micrometer
deep depression.
ents the operator with a display of the four
bores, along with an evaluation of crucial
measurement criteria. A deviation map,
highlighting the deviation of the measured
bore dimensions from the CAD model,
helps reveal micron-level surface defects.
“Regardless of the bore diameter, the
accuracy of the non-contact 3D measure-
ments is the same,” explains Bartulovic.
“Defining scanning sequences is easy and
each sequence may be saved for later recall
and execution. While the user may manip-
ulate and analyze the 3D point cloud data
in an interactive manner, the BoreInspect
system also generates automated reports.”
The system is suitable for integration into
automated continuous flow manufactur-
ing, according to Vuk. “Its compact, modu-
lar, and rugged design facilitates integration,
even in high-throughput industrial metrol-
ogy applications that would have the scanner
mounted on a stage suitable for the applica-
tion, such as a robot arm, gantry, or motor-
controlled stage, and the optical fiber con-
nection to the system’s interferometer can be
hundreds of meters long.”
Currently, Novacam’s rotational scanners
feature probes with diameters between 1
and 18 mm. Standard probes inspect bores
with diameters ranging between 6 and 28
mm, however Novacam builds custom sys-
tems for bores with diameters less than 2
mm and from 30-222 mm.
www.vision-s y stems .com
1/4/18 8:00 AM
 Machine vision: key technology for automated production.
Experience how robots react flexibly to their environment.
Meet industry visionaries and innovators, discuss important
topics such as embedded vision, and discover the path that
non-industrial machine vision is taking. At VISION, the world’s
leading trade fair for machine vision.

06 – 08 November 2018
Messe Stuttgart, Germany
www.vision-fair.de

1801vsd_11 11 1/4/18 8:00 AM

 technologytrends
continued from page 9 20% of the marine litter Bandwith (Gbits)
found was related to fishing activities. 120
“Deep-sea environments pose several imag-
Gen3 PCI Express
100
(8 lane) (63/7)
ing challenges. Capturing images in low-light,
80  
silt reflecting light back causing glare and the Thunderbolt V.3 (40/3)
reduced field of view from being under water,” 60
40 GigE
(40/30 - CAT 8)
explains Marco Boldrini, Product Manager at 40
Sony Europe Image Sensing Solutions (Wey- Gen3 PCI Express
20 ((4 lane) (31.5/7)
bridge, Surrey, United Kingdom; www.image- 10
Ca
Camera
Cam Link HS (16.8/15)
sensing-solutions.eu). “Plus, the challenge of
8 Thu
TTh
Thunderbolt V1/2 (20/3)
transmitting data to the surface from depths
U 3.1 Gen2 (10/3)
USB
6
of greater than 2,000 meters.”
CameraLink (Deca)(6.8/10)
Water, with a refractive index of 1.333, shrinks 4 (5/100)
USB 3.0 (5/3)
the field of view by 25%, so any camera chosen
2 IEEE 1394b (0.8/4.5)
needs to take this into consideration. The USB 2.0 (0.48/5)
1 (1/100)
research team acquired images using towed stills
10 20 30 40 50 60
camera moving at 0.9 m/s and with a remotely Maximum theoretical cable length (meters)
operated vehicle (ROV) running a 2008 Sony
Fiber-optic cable was required for data transmission due to distance limitations of traditional
FCB-H11 color block camera. Equipped with a
machine vision camera interface standards.
10X zoom (f1.8-2.1), the camera is capable of cap-
turing images at HD resolutions with a 50-degree Consequently, lighting needs to be carefully
horizontal viewing angle. controlled, and designed to illuminate the sea
“Light deteriorates quickly under water,” floor while avoiding blinding reflections from
explains Boldrini, “and while cameras such as silt. Indeed, the research team highlighted that
the FCB-H11 can capture down to 1 lx (and several images were discarded due to the prob-
several cameras, such as the FCB-EV7520A lem of silt creating “poor quality images due to
down to 0.35 lx), there is little significant light sediment clouds obscuring the image,” notes
found below 200 meters, and no sunlight Boldrini. “This makes it very difficult to auto-
beyond the twilight zone (200 m - 1,000 m).” mate lighting control, and even in a more con-
Sony FCB cameras help researchers monitor for pollution and coral health in the Bay of Biscay.
12 Januar y 2018 VISION SYSTEMS DESIGN
1801vsd_12 12
CoaXPress
(octal lane)
100 GigE
CXP-12.5 (100/N/A) (100/100-OM4-MMF)
(optical)
CoaXPress
(octal lane)
CXP-6 (50/68) 40 GigE
CoaXPress (40/100-OM4-MMF)
(Quad lane) (optical)
CXP-6 (25/68)
10 GigE
(10/100)
CoaXPress
(single lane)
CXP-6 (6.25/68) NBASE -T
Gigabit Ethernet
70 80 90 100 110 120
trolled system such as the ROV an operator
would ideally have the ability to calibrate lighting
and exposure times to maximize image quality.”
Another challenge is the corrosive nature of
salt water, and the pressures the systems will be
exposed to (1 additional atmosphere of pres-
sure for every 10 m in depth) means the hous-
ing used needs to be exceptionally well
designed. In this application, the camera sys-
tems must reside inside housings capable of
surviving more than 250 atmospheres of pres-
sure to operate at such depths.
“Here systems such as MacCartney Bene-
lux’s Luxus range can be used,” Boldrini notes.
“These couple Sony cameras with adaptable
lighting control and a housing capable of
reaching depths below 9,000 meters.”
Corals were found at depths of between 50 m
and 1,000 m, with the team searching for marine
litter at depths of almost 2,400 m. While an ROV
could be automated to travel at a fixed depth and
avoid canyons / objects via sonar, the challenge
of how to transmit acquired images back to the
surface for analysis in a timely manner and with-
out compromising image quality remains.
Traditional machine vision standards simply
aren’t capable of transmitting data over these
distances, Boldrini explains. “Therefor fiber-
optic cable was needed to control both the
lighting and the camera systems, but this adds
significantly to the cost of the build.”
www.vision-s y stems .com
1/4/18 8:00 AM
 technologytrends

continued from page 9 that are built in and Delamination at 20× magnification and focus
at the bottom of silicon
over the thinly sliced silicon wafer, the wafer
ultimately undergoes a multitude of micro-
fabrication process steps such as masking,
etching, doping, and metallization.
ICs, which have become the main com-
ponent of nearly all electronic devices, are
comprised of a microscopic array of elec-
tronic circuits and components that has
been implanted onto the surface of a single
crystal of semiconductor material such as
silicon. Within the integrated circuit, com-
ponents, circuits, and base materials are all Delamination at 20× magnification and focus
made together out of a single piece of sili- at the middle of silicon
con wafer. Hundreds of integrated circuits
are made at the same time on a single, thin
slice of silicon wafer and then they are cut
apart into individual IC chips.
Wafers accumulate residual stress during
the growth, sawing, grinding, etching and
polishing process, and cracks may be gener-
ated throughout all of these processes. Cracks
may also occur when the integrated circuits
themselves are cut into individual ICs. If unde-
tected, wafers can be rendered unusable in InGaAs cameras operating in the SWIR
subsequent manufacturing stages. As a result, wavelength band from 0.9 µm to 1.7 µm
acquire images that highlight defects within
it is important to inspect the raw material sub-
the silicon wafer.
strate for impurities before processing and
detect any defects during processing, in order a 30 µm pixel size and is sensitive in the
to reduce waste and keep cost down. short-wave infrared spectrum from 0.9
By using indium gallium arsenide (InGaAs) µm to 1.7 µm at up to 344 fps. The SWIR
cameras operating in the short-wave infra- camera—which is thermo-electrically
red (SWIR) wavelength band from 0.9 µm to cooled for low-noise images—has a GigE
1.7 µm, it is possible to see through semicon- Vision interface, a compact form factor (55
ductor silicon substrates. This ability to see mm x 55 mm x 78 mm), as well as Allied
through the material enables manufacturers Vision’s Vimba software development kit,
to generate infrared images that will highlight which enables users and developers to pro-
defects such as cracks within the silicon wafer. gram their own applications across Win-
Radiant Optronics—a company that has dows and Linux platforms.
a strong focus in the semiconductor field, “The Goldeye G-008 is a low-resolution
wafer fabrication, service laboratories, pack- model with an affordable price. The low
aging, and PCB assembly in Asia—has resolution is sufficient to detect the defects,
developed an infrared imaging microscope so it is the ideal choice for this cost-sensi-
that IC manufacturers can use to inspect for tive application. Our customers can bene-
internal defects and cracks. fit from the Goldeye’s outstanding perfor-
The microscope utilizes a Goldeye G-008 mance and we enjoy support from Allied
SWIR camera from Allied Vision (Stadtroda, Vision’s Asia-Pacific office just next door in
Germany; www.alliedvision.com), which Singapore,” explained Christopher Cheong,
features a 320 x 256 InGaAs sensor with Director of Radiant Optronics.

www.vision-s y stems .com VISION SYSTEMS DESIGN Januar y 2018 13

1801vsd_13 13 1/4/18 8:00 AM

1801vsd_14 14 1/4/18 8:00 AM
 Integration Insights

How deep learning is

enhancing machine vision
Developers increasingly apply deep learning and artificial neural networks
to improve object detection and classification.

“O “O
Johannes Hiltner K” K”

Good Simple
samples features

Digitalization has a firm grip on industrial pro-

duction, with processes increasingly automated Deep Complex Deep
learning features CNN
as part of the Industrial Internet of Things “N
OK
“N
OK
” ”
(IIoT). In the IIoT, which is also known as
Defective Classifier
Industry 4.0, various machines and robots take samples
on more everyday production tasks. In assembly
for example, new, compact and mobile robots,
such as collaborative robots (cobots), often work
Deep learning technologies and convolutional neural networks
hand in hand with their human colleagues.
(CNNs) can learn and distinguish between defects.
The IIoT’s highly automated and universally
networked production flows characterized by degree of automation, much greater produc- range of tasks such as fault inspection, work-
machine-to-machine interaction depend on tivity, and more reliable identification, alloca- piece positioning and the automatic handling
machine vision to reliably identify a wide range tion, and handling of a wider range of objects of objects in robotics.
of objects in the flow of goods within facto- throughout the entire value chain.
ries and the rest of the process chain. Machine As the “eye of production,” machine vision Analyze and evaluate large data sets
vision increases the efficiency and safety of these software has become an essential element of In an effort to make the identification pro-
workflows, and has become the technology, processing cess even more robust and adaptable to the
an indispensable tool for engi- unstructured data such as requirements of flexible and networked IIoT
Johannes Hiltner is a
neers seeking to automate and digital images and video gen- processes, machine vision software develop-
speed up production. product manager at MVTec erated by cameras to iden- ers increasingly rely on methods from the
Today, innovative machine Software GmbH (Munich, tify objects by their external field of artificial intelligence (AI). Deep
learning and deep learning Germany; www.mvtec. optical features alone. Such learning is an area of machine learning that
processes can ensure even software works very fast and enables computers to be trained and learn
com) and responsible for
more robust recognition rates. achieves extremely high and through architectures such as convolutional
the HALCON software, the
Thanks to such advances in reliable identification rates, neural networks (CNNs).
artificial intelligence, compa- company’s flagship product. and consequently, is used The special attribute of AI, machine-learn-
nies can benefit from a higher across industries for a wide ing and deep-learning technologies is that they

www.vision-s y stems .com VISION SYSTEMS DESIGN Januar y 2018 15

1801vsd_15 15 1/4/18 8:00 AM

 Integration Insights

to learn new things independently. By taking the

Good features of all image data into account, conclusions
can then be drawn about the properties of a cer-
? ? ? tain class, which significantly improves the iden-
tification rates. This process is called “inference.”
Therefore, deep learning algorithms are also
very suitable for optical character recognition
(OCR) applications, that is, for precisely identi-
Defective fying letter or number combinations. Due to the
extensive training process, the typical features
of the individual characters are precisely identi-
Input Output
fied based on the defined classes. However, since
there are many different fonts, some with devi-
Methods such as deep-learning technologies and convolutional neural networks (CNNs) from the
field of artificial intelligence (AI) are entering machine vision to help image-processing systems ating features such as serifs, problems may arise
learn and distinguish between defects and make identification processes even more precise. allocating them with certainty.
Advanced machine vision software can solve
comprehensively analyze and evaluate large amounts of data (big data) this problem. MERLIC and HALCON from MVTec (Munich, Ger-
in order to train many different classes and thereby more effectively dis- many; www.mvtec.com), for example, contain an OCR classifier based
tinguish between objects. Increasingly, this data is generated within the on deep learning algorithms, which can be accessed via many pre-trained
IIoT. This can be digital image information as well as data from sensors, fonts. As a result, a wide range of typefaces, such as dot-print, SEMI,
scanners, and other process components. industrial, and document-based ones, can be identified with certainty
In order to use deep learning, CNNs must first be trained. This thanks to a single, universal, pre-trained classifier.
training process relates to certain external features that are typical of
the object, such as color, shape, texture, and surface structure. The Avoid excessive training time
objects are divided into different classes based on these properties to However, companies often shy away from using AI-based technolo-
allocate them more precisely later. gies such as deep learning, since, due to their complexity, they require
In conventional machine vision methods, a developer must labori- developers to have extensive expertise. The training process generally
ously define and verify the individual features manually. With deep requires many sample images to recognize objects.
learning, however, self-learning algorithms
are used to automatically find and extract Input Input
the unique patterns in order to differentiate
Input Input
between the particular classes.

Train objects through classification Input Input

How does the training process work exactly? The
user first supplies image data that has already
Input
been provided with labels. Each label corre-
sponds to a tag that indicates the identity of the
particular object. The system analyzes this data,
and–on this basis–creates or “trains” correspond-
ing models of the objects to be identified.
Due to these self-learned object models, the
deep learning network is now able to assign
the newly-added image data to the appropriate
classes, such that, their data content or objects
are also classified. Thanks to this allocation to
certain classes, the items can then continue to
be identified automatically. Output
A sample image for direct comparison is
therefore no longer necessary for each individual Modern machine-vision solutions enable companies to train neural networks themselves. Photo
object. After all, deep learning processes are able credit: MVTec Software GmbH

16 Januar y 2018 VISION SYSTEMS DESIGN www.vision-s y stems .com

1801vsd_16 16 1/4/18 8:00 AM

 Integration Insights

Up to 100,000 comparison images may be needed for each class
in order to achieve adequate recognition rates. Even if the necessary
sample data is available, the training process takes up an enormous
amount of time. Usually, the programming work for identifying dif-
ferent defect classes during fault inspection is extremely complex, too.
The reason is that highly skilled employees with suitable training are
required for this purpose.
Modern machine vision solutions, which already include a large
number of deep learning functions, can help. The new version 17.12 of
the standard software MVTec HALCON enables companies to train
convolutional neural networks (CNN) themselves without a great deal
of time and money. After all, the software is already equipped with two
networks that are optimally pre-trained for industrial use — one is opti-
mized for speed and the other for maximum recognition rates.
The training process, therefore, works with only a few sample images
provided by the customer, and thus are tailored to the customers’ exact
applications, resulting in neural networks that can be precisely matched
to the customer’s specific requirements.
User companies can significantly reduce the amount of programming
work needed by easily and systematically classifying new image data,
saving time and money. Normally, they do not have to have any in-depth
AI expertise. Companies can use their existing personnel without prob-
lems to train the network.
and other components are reliably identified, which allows the removal
of corresponding parts to be automated.
The food and beverage industry benefits from deep learning tech-
nologies, too. For example, poor-quality fruits and vegetables can be
detected more precisely before they are packaged or further processed.
The processes are also used in automotive engineering. This indus-
try, in particular, is characterized by an especially high degree of auto-
mation. Here, for example, self-learning algorithms perfectly identify
tiny paint defects that are not visible to the naked eye.
Another important area of application is pharmaceuticals. Pills often
look very similar on the outside, but contain entirely different active sub-
stances. Through deep learning and CNNs, the drugs can be very reli-
ably identified, inspected, and distinguished from each other so they are
always placed in the correct blister packs.
Conclusion
Technologies based on artificial intelligence, such as deep learning and
CNNs, are an important part of modern machine vision solutions today.
Deep learning enables companies to train neural networks themselves
without any in-depth expert knowledge and minimal effort, especially
when programming defect classes during error inspections. The result
is that companies can save money and benefit from much more robust
recognition rates as well as better classification results.

TEN
Detect defects efficiently
Recognizing defects is a time-consuming process because the appear-
ance of defects, such as tiny scratches on an electronic part, can never
be accurately described in advance. Therefore, it is very difficult to man-

YEARS
ually develop suitable algorithms that can detect any conceivable faults,
based on sample images. An expert would have to manually view hun-
dreds of thousands of images and program an algorithm that describes
the error as precisely as possible based on his observations. This would
SINCE
SI
INCE
NCE ITS NTRO
ITS INTR
IINTRO
simply take too long.
AND STILL

GOING
Deep learning technologies and CNNs, on the other hand, can inde-
pendently learn certain characteristics of defects, and precisely define
the corresponding problem classes. So, only 500 sample images are
needed for each class, based on which the technology trains, verifies,
and thereby precisely detects the different types of defects.
This process takes only a few hours. Not only does it minimize the
STRONG
amount of time required, but the recognition rates are also much higher
For the ultimate in flex life performance
rmance
than with manually programmed defect classes. The self-learning algo-
demand Intercon Infiniflex assemblies
mblies
rithms, therefore, help to significantly reduce identification errors, while featuring GORE® high flex cable.
the error quotas for manual programming can be inefficiently high. Supporting all of the newest protocols
tocols
including Ethernet, Camera Link, USB, Featuring
GORE CameraLink
® ®

Many industries benefit CoaXPress and more. High Flex Flat Cables

Machine vision technologies based on deep learning and CNNs can be

used profitably in many different branches of industry and applications.
Precision Cable
In the electronics industry, the inspection process can be automated and Assemblies for the
Vision and High Speed
accelerated. With the help of self-learning methods, therefore, all con- Manufacturing Industries

ceivable product defects can be effectively detected – as described above.

(218) 828-3157 | intercon@nortechsys.com | www.intercon-1.com
Even the tiniest scratches or cracks in circuit boards, semiconductors,

www.vision-s y stems .com VISION SYSTEMS DESIGN Januar y 2018 17

1801vsd_17 17 1/4/18 8:00 AM

 Integration Insights

When to invest in LED

strobe controllers
LED strobe controllers offer many benefits for applications involving low
contrast, motion, mark quality assessment, OCR and ambient lighting issues.

Alan L. Lockard

After many years as an end user designing many new LEDs, as they heat up from 25°C Forward current
IF = f(VF ); TA = 25°C
and installing machine vision inspection sys- – 90°C, the brightness of the LEDs can drop 100
tems without adding a strobe controller, it’s by up to as much as 40%. A
IF
has become clear that the benefits of LED This reduction in output intensity may
lighting controllers far outweigh the cost in begin manifesting itself by means of false
applications that involve: rejects. Eventually, adjustments to either the
• Low contrast lighting output or the camera system will be
• Inspection and motion required to restore optimal performance.
• Bar code grading and mark A quality LED lighting controller can pro- 10-1

quality assessment vide consistently-uniform lighting over the

• Optical character recognition life of the project. For continuous operation,
• Ambient lighting issues the life of the LED array can be extended
• Low contrast by operating at less than 100% power. For
Machine vision applications involving strobe control, even when overdriving at
image contrasts below 20% present specific lower duty cycle operation, the LED array 10-2
2.0 2.5 3.0 3.5 4.0 4.5 5.0 V 6.0
challenges. In such low contrast applications, will last longer. VF
Typical LED output curve
even the slightest variation in LED output Operating the LED array at reduced
intensity can have a substantial impact on output or in strobed mode reduces heat gen- Figure 1: While LED lights are specified as
the performance of the eration and substantially either 12V or 24V, the actual light output from
inspection system. Alan L. Lockard is an Ad- slows the degradation of the LED results from current through the semi-
conductor device. Consequently, all LED device
An LED array operat- vanced Certified Vision Profes- LED array output. LED
manufacturers specify current control for the
ing at 100% power gen- controllers that regulate
sional and Principal Engineer in most efficient use.
erates heat that short- the current to the LED
ens the life of the LED Reagent Engineering at bioMéri- array rather than the Inspection and motion
array, gradually dimin- eux, Inc. (Hazelwood, MO, USA; voltage also provide more In-line machine vision systems tend to be
ishing the output inten- www.biomerieux.com). consistent illumination installed on machines where the object being
sity over time. Even with intensity (Figure 1). inspected is moving on a conveyor. Depending

18 Januar y 2018 VISION SYSTEMS DESIGN www.vision-s y stems .com

1801vsd_18 18 1/4/18 8:00 AM

 on the speed of the conveyor, the image will
be blurred due to the amount of movement of
the object during the exposure period. Factors
involved are time (exposure), area (aperture)
and intensity (light).
If time is reduced by use of a shorter expo-
sure time to mitigate blur, then either the area
or the intensity must be increased for sufficient
exposure. While opening the aperture by one
f stop doubles the amount of light gathered
by the lens, it also reduces the depth of field
(DOF) which may be undesirable.
Continuous lighting is limited to what light
output is available at 100% power. The only
way to increase intensity of continuous light-
ing is to increase the size and wattage of the
LED array. This also increases initial expense,
operating costs and produces undesirable heat.
Pulsing or strobing the LED array, espe-
cially overdriving it, provides high intensity
illumination at short duty cycle. This comple-
ments short exposure imaging used for moving
objects, and overdriving the array has no detri-
mental impact on the lifespan or performance
of the array (Figure 2).
Barcode grading (verifying) and OCR
Inspection of barcodes at the point of printing
helps to ensure reliable readability in the field.
Referred to as verifying or grading, there are a
variety of recognized standards in use today.
All compare numerous physical attributes,
each scored individually, and combined into
an overall print quality grade.
Attributes such as symbol contrast and mod-
ulation are affected by illumination intensity,
exposure time and system gain. Once set,
exposure time and system gain remain rela-
tively constant in the machine vision camera.
However, because LED output degrades over
time, only a high-quality LED lighting con-
troller can ensure consistently uniform light-
ing for the life of the project in such barcode
verification applications.
Mark quality assessment at the point of
printing the barcode is likely to be on a moving
object or surface. Therefore, the steps to miti-
gate blurring of the moving image mentioned
earlier must be considered. Stop action imag-
ing, by combining short exposure time and
high intensity strobing, will provide consistent
www.vision-s y stems .com VISION SYSTEMS DESIGN
1801vsd_19 19
Exposure Exposure
begins ends
Camera Exposure
Illustraton Minimum 5x
intensity with
pulsed supply
Overdriven pulse
Light intensity
100% output
with continuous
supply
Continuous intensity lighting
Lighting trigger
Exposure trigger
Figure 2: Converting a constant illumination system to pulse illumination is straightforward. The
trigger for the camera is sent to a lighting controller. The controller provides precise pulse width
timing, power and brightness control for the lighting pulse. This ensures that the lighting pulses
during the camera exposure time and that the light energy is the same for every image.
barcode verifying without motion blur that margin for print quality degradation caused
could result in poorer barcode grades. Cali- by subsequent handling and packaging.
bration of the machine vision system for bar The principles just discussed for barcode
code grading can be done using a Calibrated grading also apply to optical character recog-
Conformance Standard Test Card. nition (OCR). Selecting a quality LED light-
One application involves a GS1 DataMa- ing controller and using OCR fonts when
trix barcode being printed and verified on web- printing provide an extremely robust OCR
bing moving at 750 inches per minute. The reading application.
machine vision exposure was fixed at 250µsec Many smart camera-based machine vision
to minimize motion blur. Because the GS1- systems come with auto-tuned OCR capabili-
specified minimum resolution is 10 pixels per ties. These read consistently printed and illu-
cell, the working distance was set so that the minated images of OCR fonts with a high
resolution was approximately 10 pixels per cell. degree of success. Using a quality LED strobe
The LED array was strobed for 1msec at controller enables consistently reading of 2mm
500%, and the gain of the machine vision high, low contrast, laser marked, OCR font
sensor was set to 1, then increased incremen- characters with a high degree of reliability.
tally until the Symbol Contrast (SC) measured
by the sensor matched the SC recorded on the Ambient lighting
test target. Making gain adjustments from the For a machine vision system to work predict-
bottom up proved more reliable. ably and accurately on the plant floor, there
This methodology produced an in-line ver- should be no effects due to ambient lighting.
ifier that grades GS1 barcodes in harmony If an inspection is influenced by the operator
with the benchtop verifier being used by the standing between the overhead lighting and
QC department to verify QC samples. Set- the inspection, there will be problems.
ting the acceptance criteria on the in-line One method of mitigating the effect of
verifier one letter grade higher than that of ambient lighting is to provide some sort of
the QC benchtop verifier provides a safety shrouding over the inspection area to block out
Januar y 2018 19
1/4/18 8:00 AM
 NEW PRODUCTS –
NEW POSSIBILITIES
a)
Telecentric
lenses with
tunable working
distance
ambient light. However, such an approach
also obstructs visual inspection of the object.
In pharma and other quality-critical envi-
ronments, shrouding sections of the produc-
tion line poses issues with line-clearance pro- b)

cedures. By using high intensity strobing, the

 focusing without influence of ambient lighting on the inspec-
moving elements tion can be substantially reduced.
 0.13x - 0.66x for sensor With an overdriven LED strobe, the ambi-
up to 16 mm
ent light is literally overpowered. This theory
 1x - 3x for sensors
up to 35 mm can be easily tested by unplugging the LED
array and triggering an inspection. If the
inspection image is completely dark, the ambi-
Figure 3: The Gardasoft RT220-20 two chan-
nel (left) and RC-120 single channel (right) con-
trollers regulate current to LED lights, provide
overdriving capability and are only configu-
rable over Ethernet via embedded web pages
to help keep validated settings in their validat-
ed state by limiting access.

ent light is very unlikely to have an impact on

Telecentric
lenses with
integrated coaxial
illumination
 improved image homo-
geneity and intensity
 exchangable beamsplitter
(unpolarized, polarized)
 possibility to integrate
a retardation plate
Visit us at
South hall, booth 1146
SILL OPTICS GmbH & Co. KG
Johann-Hoellfritsch-Str. 13
DE-90530 Wendelstein
Phone: +49 (0)9129 - 90 23-0
info@silloptics.de • silloptics.de
the inspection.
When using monochromatic machine
vision cameras, use of monochromatic LED
lighting can also help mitigate the effects of
ambient lighting. Choosing the appropriate
monochrome LED color is really a matter of
what works best with your inspection appli-
cation. However, once a color is chosen, use
of a matching bandpass color filter on the
machine vision camera lens limits the impact
of the ambient lighting to only the range of the
bandpass filter.
Use of a monochromatic LED light source
provides a narrow but repeatable imaging sce-
nario for the machine vision system. Reducing as
many extrinsic variables of a machine vision
system adds to system performance and reliability.
Choice of LED controller
Because these applications were all validated
systems, RT220-20 two channel and RC-120
single-channel controllers with Ethernet com-
munications from Gardasoft Vision (Cam-
bridge, UK; www.gardasoft.com) were speci-
fied (Figure 3). These controllers regulate the
current to the LED light array, provide over-
driving capability and are only configurable
over Ethernet via embedded web pages, which
keeps the validated settings in their validated
state by limiting access.
Snippets of program code included with
the machine vision smart cameras used pro-
vide communication with both the RT and
RC controllers. For new “Greenfield Appli-
cations” it would be recommended to couple
Gardasoft controllers and machine vision cam-
eras that utilize triniti technology. This tech-
nology offers tight integration of the machine
vision camera and the LED controller.
Whereas the applications mentioned in this
article are fixed and unlikely to require recon-
figuration over the life of the system, many
machine vision applications become part
of more flexible automation requiring a rec-
ipe-based solution for many different product
families. Having such an integrated machine
vision camera and lighting configuration will
help to meet such demanding applications
with greater ease than previous platforms.
This article does not reflect the views or position
of the Company but those of the author.

20 Januar y 2018 VISION SYSTEMS DESIGN www.vision-s y stems .com

1801vsd_20 20 1/4/18 8:00 AM

 Industry Solutions Profile

Embedded camera
brings 3D mapping
to boring machines
Mounting an embedded camera within the cutting head of a tunnel boring machine
allows contractors to document rock face structures as boring progresses.

Robert Wenighofer

In the past, drilling and blasting methods were

used to excavate tunnels, a process that was
labor intensive and time consuming. Today,
more efficient tunnel boring machines
(TBMs) are used as an alternative, produc-
ing smooth tunnel walls while limiting any
disturbances to the surrounding soil or rock.
Although expensive, such TBMs have the
advantage over drilling and blasting meth-
Figure 1: Multiple disc cutters are used in a tunnel boring machine to cut the rock-face. In a
ods since, when tunnel lengths are long, they
single disc housing, the disc cutter (middle right) shown in this front elevation, rotates as it cuts
promise to be more efficient and can perform the rockface. By housing a GigE camera and control unit (shown in red) within this disc cutter
such tasks in a relatively short period of time. (top, left), a circular 85° field of view (FOV) of the face can be imaged as the TBM rotates.
One of the advantages of drilling and blast-
ing methods, however, is that open access is unileoben.ac.at) has developed an embedded trol unit mounted in the disc cutter housing
provided to the tunnel face, making it rela- camera system to provide 3D reconstructions of the TBM (Figure 1). In a single disc hous-
tively simple to observe, analyze and record of the tunnel face. In operation, the system ing, the disc cutter (middle right), shown in
geological features. Such geological mapping visualizes the spatial position of stratifications, front elevation, rotates as it cuts the rockface.
is more difficult when using mechanized tun- the depth of any cavities and any discontinui- Currently, the TBM is building the explor-
neling methods, since the ties of the rock. atory tunnel of the Brenner Base Tunnel,
cutting head obstructs the Robert Wenighofer, Developed in collabora- the world’s longest tunnel now under con-
view of the tunnel face. To The Institute for Subsurface tion with Geodata Group struction. The camera system was deployed
overcome this, the Institute (Leoben, Austria, www. at Tulfes-Pfons section of the tunnel at Inns-
Engineering at the Monta-
for Subsurface Engineering geodata.at), the system bruck, Western Austria by a joint venture
at the Montanuniversität nuniversität Leoben (Leoben, comprises an off-the-shelf between Strabag AG (Cologne, Germany;
Leoben (Leoben, Austria; Austria; www.subsurface.at) camera mounted in a rug- www.strabag.com) and Salini Impregilo S.p.A
www.subsurface.at, www. gedized housing and a con- (Milan, Italy; www.salini-impregilo.com).

www.vision-s y stems .com VISION SYSTEMS DESIGN Januar y 2018 21

1801vsd_21 21 1/4/18 8:00 AM

Industry Solutions Profile

Rotation

After cutting, it is necessary to visualize any
new geological features that may be pres-
ent. To do so, the boring machine is
retracted and the material between
the cutter head and face cleaned. A
GigE camera (shown in red) and
control unit housed within one
of the disc cutters (top, left) then
images a circular 85° field of
view (FOV) of the face as the
TBM is manually rotated away
from the rockface.
To expand this coverage, the
same camera system is then re-
positioned in a different disc
housing, and the TBM manu-
ally rotated again. This process
is then repeated at four or five
different disc housing positions
approximately four or five times
(Figure 2). While the TBM head
rotation takes approximately 3 mins,
it takes 10 to 15 mins to then reposition
the camera system at each different disc
housing location. To expedite this process,
it would at first appear that several embed-
ded cameras could be used. However, the
massive 25-ton force that a single disc cutter
exerts and the tons of excavation material pro-
duced would impede permanently mounting
a camera system in the cutter head.
Camera and controller
To image the rockface as the TBM head is man-
ually rotated, a Prosilica GT2000 GigE camera
from Allied Vision (Stadtroda, Germany; www.
alliedvision.com) with a 2/3in CMOS 2048 ×
1088 sensor from ams Sensors Belgium (Ant-
werp, Belgium; www.cmosis.com) was fitted
with a 5mm focal length C-Mount lens to
provide the 85° FOV of the face. The camera
system is housed in a ruggedized housing incor-
porating a white LED ring light developed by
Geodata Group (Leoben, Austria; www.geo-
data.at). While the camera is operated in a con-
tinuous mode, capturing two frames per second,
the white LED is strobed over a period of 4ms to
ensure a visible light output of 10,000 Lumens so
that no motion blur occurs as the camera system
is rotated. This level of illumination is sufficient
to illuminate break-outs in the rockface of more
Figure 2: Cutter head of the TBM. To obtain a com-
plete image of the rockface, the camera is placed
at either four (shown in green) or five posi-
tions (shown in red) within disc housings of
the cutting head while the camera control-
ler is constantly mounted in a double disc
cutter near the rotation axis of the TBM.
This results in between either 700 or
1100 images being captured depend-
ing on whether four or five positions
are employed.
the camera at 2 fps which provides
sufficient overlap between images
to perform 3D reconstruction and a
surplus of images. To ensure that the
position of each image is known as
the TBM head rotates, the camera
control unit incorporates a uniaxial
inclination sensor from Posital Fraba
(Heerlen, The Netherlands; www.pos-
ital.com) mounted inside the camera
controller and interfaced to the embed-
ded PC over an RS-232 interface.
In this way, as each image is captured by
the camera, it is assigned an angular value from
the inclination sensor. This inclination angle
and the known relative position of the disc hous-
ing is then used to determine the absolute posi-
than 1.7 m depth as can be seen in Figure 4. tion of the camera in the 3D coordinate system
As images are captured by the GT2000 of the cutter head. Knowing this information,
camera, they are transferred over its GigE both two-dimensional and 3D photogrammet-
interface to a fanless quad-core embedded PC ric data of the rockface can then be generated.
from Vecow (New Taipei City, Taiwan; www. It is necessary to visualize the quality of the
vecow.com) that acts as a camera control unit. images of the tunnel face before the TBM
Camera control software running on the PC, is rotated to ensure that the image quality is
based on Allied Vision’s Vimba SDK, triggers acceptable and that the protective glass of the
Companies mentioned
Agisoft Geodata Group Strabag AG
St. Petersburg, Russia Leoben, Austria Cologne, Germany
www.agisoft.com www.geodata.at www.strabag.com
Allied Vision NVIDIA The Institute
Stadtroda, Germany Santa Clara, CA, USA for Subsurface
www.alliedvision.com www.nvidia.com Engineering at the
Montanuniversität
ams Sensors Posital Fraba Leoben
Belgium Heerlen, The Nether- Leoben, Austria
Antwerp, Belgium lands www.unileoben.ac.at
www.cmosis.com www.posital.com
Vecow
Autodesk Salini Impregilo New Taipei City, Taiwan
San Rafael, CA, USA S.p.A www.vecow.com
www.autodesk.com Milan, Italy
www.salini-impregilo.com

22 Januar y 2018 VISION SYSTEMS DESIGN www.vision-s y stems .com

1801vsd_22 22 1/4/18 8:00 AM


camera unit is not fogged up or soiled. Because
of this, the camera controller is set to automat-
ically initiate recording after the camera con-
trol unit has powered up. Then, by incorpo-
rating a wireless network card in the camera
controller, stationary images can be analyzed
remotely using any Android-powered hand-
held device such as a smartphone or a personal
digital assistant (PDA).
Mapping the face
To obtain a complete image of the rockface,
only the camera unit is placed at either four
(shown in green) or five positions (shown in
red) within the disc housings of the cutting
head. This results in between either 700 or 1100
images being captured depending on whether
four or five positions are employed. After each
image is captured, it is stored along with is asso-
ciated unique angular value from the inclina-
tion sensor. After these images are captured, the
camera and camera controller are removed and
the data transferred over a USB interface to an
Intel i7-based desktop PC with a GTX970 PCI
Express graphics card from NVIDIA (Santa
Clara, CA, USA; www.nvidia.com).
To generate 2D and 3D maps of the rockface
and allow image measurements to be made,
captured images are then processed using the
Photoscan software package from Agisoft (St.
Petersburg, Russia; www.agisoft.com). Using
this software, the camera can be first cali-
brated before use to rectify any possible distor-
tion from the camera’s lens as well as comput-
ing the parameters of the camera matrix such as
the focal length and projection center of the lens.
This information along with each 2D image and
camera position data from the inclination sensor
is then processed by Photoscan in a process
known as bundle block adjustment. In this pro-
cess, a set of 2D images are taken from different
viewpoints and the relative motion of the camera
and its optical characteristics are used to gener-
ate an optimal 3D reconstruction of the scene.
Using the 2D image sequences generated
as tunneling progresses, geologists can char-
acterize and measure such features as spatial
discontinuities and visualize the behavior of
the excavation process (Figure 3). In this exam-
ple, three images of the rock face are shown
as tunneling progresses (from left to right). In
www.vision-s y stems .com VISION SYSTEMS DESIGN
1801vsd_23 23
Industry Solutions Profile
-0.60m
akk
brreea
Foliation
OOuuttb
Harnisch
Drive
direction
Drive direction
-0.30m
Figure 3: Using the 2D image sequences generated as tunneling progresses, geologists can char-
acterize and measure such features as spatial discontinuities and visualize the tunnel face and the
behavior of the excavation process as it continues
-0.70m
-0.30m
Figure 4: Color-coded elevation images that represent the distance between the cutting head
and the tunnel face can also be used to detail the structure of the surface of the face. In this col-
or-coded image of the rockface (left), a huge break-out can be clearly visualized.
the monochrome image on the left, there is a repair or replacement to the cutting discs. For
failure of the tunnel face which may be due contractors to objectively document this is vital
to structural weakness in the rock. To further since any such possible repairs may be expen-
analyze this structural weakness geological fea- sive and need to be passed on to their customers.
tures like schistosity can be displayed (graphic, Field-tested in the construction of the
far left) as a stereographic projection. This is Brenner Base Tunnel in the Austrian Alps, the
accomplished using a software plug-in based camera system has been used for more than
on the .NET application programming inter- a year where it has proven to be an appropri-
face (API) of AutoCAD software from Autodesk ate means of documenting and visualizing the
(San Rafael, CA, USA; www.autodesk.com). rockface even in the harsh environments cre-
Color-coded elevation images that represent ated by TBMs. Patent protection for the system
the distance between the cutting head and the is currently being sought in Middle-Europe.
tunnel face can also be used to detail the struc-
ture of the surface of the face (Figure 4). In this Reference:
color-coded image of the rockface (left), dis- Wenighofer, R., Galler., R., Six, G., Chmelina,
continuities can be clearly visualized. This is K.: Cameras for the digitized 3d geological
important since, when tunneling, the cutting documentation of the tunnel face from cutter
disks will be subject to different forces at differ- heads of TBMs, Geomechanics and tunneling,
ent points on the surface possibly requiring later 10 (2017) No. 6, 760-766
Januar y 2018 23
1/4/18 8:00 AM
 High Resolution
graphics: www.donaldecompany.com

meets Speed

• High Speed CoaXPress Machine Vision cameras • Easy to use high speed recording software:
• High Resolution Machine Vision cameras Visual Marc
• Small ruggedized high speed recording cameras • Long term longevity Vision PCs
• Long time recording systems • Custom camera solutions

Visit us at Table/Booth 54 - South Lower Lobby | San Francisco CA, USA | 30 Jan - 1 Feb 2018
www.mikrotron.de

Member of
www.lakesighttechnologies.com

1801vsd_24 24 1/4/18 8:00 AM

 product focus on High-speed imaging
High-speed cameras target
machine vision applications
Faced with imaging high-speed events, developers can choose from those that transfer
data over industry-standard camera-to-computer interfaces and stand-alone cameras.
Andrew Wilson, European Editor
Traditionally, high-speed cameras were
expensive and relegated to applications such
as ballistics applications and automotive test-
ing where the cost of such systems was second-
ary. Today, however, with the advent of high-
speed CMOS imagers, low-cost memory,
high-speed camera interfaces and low-cost
image capture and analysis software; high-
speed cameras have found wider market
acceptance in fields such as scientific research
and film and video production.
CCD vs CMOS
In the late 1970s, the emergence of CCD cam-
eras began to challenge film-based products
and numerous CCD architectures including
full-frame, frame-transfer and interline-trans-
fer devices were devised to meet the demands
of different applications. For high-speed
imaging, the disadvantages of using CCDs
included the need to employ mechanical shut-
ters with full-frame devices, charge smearing
associated with frame transfer devices and the
reduction in sensitivity of interline-transfer
devices due to the devices’ interline mask (see
“Architectures commonly used for high per-
formance cameras,” http://bit.ly/VSD-CCD,
on the website of Andor Technology (Belfast,
Ireland; www.andor.com).
In the mid-1990s, CMOS-based imag-
ers began to challenge the stronghold once
commanded by CCD-based devices. These
devices have several advantages over CCD-
based sensors when used in high-speed
www.vision-s y stems .com VISION SYSTE M S DESIGN
1801vsd_25 25
camera systems, namely faster speeds
and lower power consumption. They
have gained acceptance in high-speed
camera designs because by individu-
ally digitizing the captured charge at
each photosite, blooming and smear-
ing is eliminated and, when oper-
ated in global shutter mode, each pixel is
exposured simultaneously, eliminating spa-
tial aberrations from extremely fast-moving
objects. Such global shutter modes are, how-
ever, achieved at the cost of a significantly
reduced frame rate over that of operation
in rolling shutter mode (see “Rolling Shut-
ter vs. Global Shutter,” http://bit.ly/VSD-
SHUT, on the website of QImaging (Surrey,
BC, Canada; qimaging.com).
Figure 1: Optronis’ CR-S3500, is a 1280 x 860
CMOS-based camera capable of running at
3,500 fps at full pixel count and specified as
having an adjustable exposure time from 2μs
to 1/frame rate, i.e. 2-286μs.
Figure 2: The IL5, a 2560 x 2048 CMOS-
based camera from Fastec Imaging with a max-
imum frame rate of 253 fps at full pixel count
can be operated in a number of window-
ing modes. The fastest of these windows the
imager to 64 x 32 pixels allowing a frame rate
of 29,090 fps to be achieved.
Image motion
Knowing the exposure times of a high-speed
camera is very important since this, along
with the velocity of the moving part, the
number of pixels in the image sensor and
the camera’s field of view (FOV) can have
a dramatic impact on the pixel blur that
occurs when an image is captured.
Needless to say, to make accu-
rate image measurements, the
less this pixel blur is, the more
accurate the data analysis.
As Perry West, President of Auto-
mated Vision Systems (San Jose, CA,
USA; www.autovis.com) points out in
his excellent paper “High Speed Real-Time
Machine Vision,” http://bit.ly/VSD-HS, the
magnitude of the pixel blur can be mathemat-
Januar y 2018 25
1/4/18 8:00 AM
 product focus on High-speed imaging

ically described as: B = (Vp x Te x Np)/FOV
where Vp is the velocity of the moving part, Te is
the exposure time in seconds, Np is the number
of pixels spanning the view and FOV is the
field of view. Thus, for a part moving at 30cm/s
across a 1280 pixel imager over a camera FOV
of 1000cm, an exposure time of 33ms (0.033s)
would result in a pixel blur of 1.26 pixels. Even
a one pixel blur such as this, Perry suggests, may
become an issue in sub-pixel precision applica-
tions but one that can be resolved by decreas-
ing the exposure time of the camera. Unfortu-
nately, for those choosing a high-speed camera,
many manufacturers do not specify the expo-
sure times that can be achieved, merely specify-
ing the frame rate of the camera. Indeed, some
even state that the exposure time achievable is
the reciprocal of the frame rate (which it may
be in some cases).
Frame rate relates to how many frames
are captured each second and shutter
speed specifies how long each individual
frame is exposed (which can vary). For a
graphical explanation of frame rate and
exposure, see “Shutter Speed vs Frame
Rate,” http://bit.ly/VSD-SSFR. Thus,
companies such as Optronis (Kehl, Germany;
www.optronis.com) in its CR-S3500, a 1280 x
860 CMOS-based camera capable of running
at 3,500 fps at full pixel count, is specified as
having an adjustable exposure time from 2μs
to 1/frame rate, i.e. 2-286μs (Figure 1).
Regions of interest
Unlike CCD sensors that, in partial scan mode,
may suppress entire lines in the vertical direc-
tion, CMOS sensors can be operated in region
of interest mode (ROI) where windowed ROIs
Figure 3: Emergent Vision Technologies
HR-2000 is a CMOS-based camera capable of
running 2048 x 1088-pixel images at 338 fps.
In windowing mode, this can be reduced to 320
x 240 to achieve a 1471 fps rate. With a band-
width of 10Gbits/s, images can be transferred
to a PC without requiring host camera memory.
in both the horizontal and vertical directions
across the sensor can reduce the image size and
thus increase the readout speed.
For this reason, many manufacturers of
high-speed cameras specify both the maxi-

Understanding ISO in a digital world

Many of those responsible for specifying high-speed cameras will real-
ize the importance of sensitivity. Those in the machine vision industry
realize that the specific sensitivity of any given camera is device spe-
cific and depends on a number of factors including the quantum effi-
ciency, pixel size, and the shot noise and temporal dark noise associated
with the CCD or CMOS imager used in the camera (see “How to Evalu-
ate Camera Sensitivity,” FLIR White Paper; http://bit.ly/VSD-CAMSEN).
For those involved in high-speed imaging, the ISO standard is often
used to describe sensitivity. ISO Sensitivity (or ISO speed) is a measure of
how strongly an image sensor or camera responds to light; the higher the
sensitivity the less light that is required to capture a good quality image.
However, the ideas and measurements behind it are less well known. If
two cameras are rated at ISO 1200, will they produce the same images
for the same amount of light? Regrettably, the answer is “not necessarily.”
Film sensitivity measurements began in the late 1800s and, since
then, many organizations have vied to produce the dominant standard.
DIN and ASA ratings were the de-facto standards for many years and, in
1974, the International Standards Organisation (ISO) started collecting
these together, eventually creating ISO 6, 2240 and 5800.
In 1998, as digital cameras became ubiquitous, ISO created a new
standard specifically for digital still cameras. The latest version, ISO 12232-
2:2006 has become the de-facto standard for digital still and video cam-
eras. The rigorous method in the standard requires an illuminated scene,
a camera to collect images and a measurement or assessment of the
images. Unfortunately, there are options in each of these steps.
For example, the standard allows the use of either daylight or tungsten
lighting. For a monochrome camera, especially one without an IR-cut filter,
tungsten illumination is advantageous. This is supposed to be declared with
a “T”, such as ISO 1200T. However, the use of a “D” for daylight is optional,
so the compulsory “T” could be lost without being noticed.
Also, illumination in the test can be measured at the scene as either a
“scene luminance method” or at the sensor as a “focal plane method.”
The mathematics in the standard should yield the same value for both
techniques, but there is a temptation to try them both to see if one
gives better results.
The biggest discrepancies come from the choice of image rating
technique. There are two noise-based speed measurements, Snoise10
and Snoise40, which are related to film standards. There is also the sat-
uration-based speed measurement, Ssat, but this method does not pre-
vent manufacturers from using an undisclosed amount of gain. The con-
cept of recommended Exposure Index (EI) correctly allows for gain in the
camera, but this is closely related to another measurement, the Stan-
dard Output Sensitivity whose result does not mention gain. In truth,
gain is not necessarily bad, but it will increase noise in an image, which
may prove more undesirable than lower sensitivity.
The differences between Saturation-based and Standard Output
Sensitivity are documented in “ISO Sensitivity and Exposure Index,”
- Technical Documentation from Imatest (Boulder, CO, USA; www.
imatest.com) that can be found at http://bit.ly/VSD-IMA.
Choosing a camera would be easier if all manufacturers were to use
the same measurement - perhaps an update to the standard would be
beneficial. In the meantime, the saturation method, with true disclosure
of gain and light source technology would seem to be a good baseline.
Chris Robinson, Director of Technology, iX Cameras (Woburn, MA, USA;
www.ix-cameras.com)

26 Januar y 2018 VISION SYSTEMS DESIGN www.vision-s y stems .com

1801vsd_26 26 1/4/18 8:00 AM

 mum frame rate and the ROI used to achieve
these rates. Fastec Imaging (San Diego, CA,
USA; www.fastecimaging.com), for example,
in the design of its IL5, has developed a 2560
x 2048 CMOS-based camera (Figure 2) with
a maximum frame rate of 253 fps at full pixel
count, that can be operated in a number of
windowing modes. The fastest of these win-
dows the imager to 64 x 32 pixels allowing
a frame rate of 29,090 fps to be achieved.
With the large amount of data being cap-
tured by such cameras, systems develop-
ers need to consider whether the current
camera-to-computer inter-
faces are enough to sustain
high-speed image data trans-
fer to a host computer. If so,
then a number of high-speed
interfaces are available with
which to implement such systems including
10GigE, USB, CoaXPress and less-commer-
cially popular camera-to-computer interfaces
such as Thunderbolt and PCI Express. If not,
then such cameras must be equipped with
on-board image memory to store sequences of
high-speed images that can then be later trans-
ferred to host computers for further analysis.
Two examples highlight the image transfer
speeds that can be achieved with the latest cam-
era-to-computer interfaces. Using the 10GigE
interface standard, Emergent Vision Technol-
ogies (EVT; Maple Ridge, BC; Canada; www.
emergentvisiontec.com), for example, offers its
HR-2000, a CMOS-based camera capable of
running 2048 x 1088 pixel images at 338 fps
(Figure 3). In windowing mode, this can be
reduced to 320 x 240 to achieve a 1471 fps rate.
With a bandwidth of 10Gbits/s, this is ample
enough to transfer such images to a PC with-
out requiring host camera memory.
Similarly, in its EoSens 25CXP+ 5120 x 5120
CMOS-based camera, Mikrotron (Unterschleis-
sheim, Germany; www.mikrotron.de) has imple-
mented a four-channel CoaXPress implementa-
tion capable of transferring data at 25Gbits/s. In
windowing mode, the 80 fps rate of the camera’s
full 5120 x 5120 pixel imager can be reduced to
1024 x 768, increasing the frame rate to 423 fps.
Other high-speed interfaces such as the
10Gbit/s Thunderbolt 2 and the 64Gbits/s
PCIe Gen3 interfaces have been employed by
www.vision-s y stems .com VISION SYSTEMS DESIGN
1801vsd_27 27
Ximea (Münster, Germany; www.ximea.com)
in the MT023MG-5Y and xiB-64 series of cam-
eras respectively. Forthcoming standards such
as USB 3.2 Gen 2 also promise faster data trans-
fer rates for developers of high-speed cameras.
Although companies such as FLIR Integrated
Imaging Solutions (Richmond, BC, Canada;
COMPACT
C-MOUNT
LENSES
• Equipped with our original
Schneider-Kreuznach
Figure 4: The Phantom V2512 camera from
focusing mount
Vision Research is based around a custom
1280 x 800 CMOS imager that is capable
• Stable even in harsh
of capturing 12-bit pixels at speeds of up to
25,000 fps. This equates to an image capture environmental conditions
data rate of approximately 307Gbit/s.
• Proven since 1993
www.flir.com/mv) and others have not, at the
time of writing, offered any products based on Quality, Lightweight
this standard, the company does offer a number & Cost-Effective
of products based on the 5Gbits/s USB 3.1 Gen
1 standard and is likely to offer 10Gbit/s USB
3.2 Gen 2 cameras soon.
Going faster
While high-speed interfaces are allowing rel-
atively high-speed data transfer between cam-
eras and host computers, even the fastest
implementations cannot be used for some of
the most demanding high-speed applications.
For example, the Phantom V2512 camera
ENGINEERED
from Vision Research (Wayne, NJ, USA; www. IN GERMANY
phantomhighspeed.com) is based around a
custom 1280 x 800 CMOS imager that is capa-
ble of capturing 12-bit pixels at speeds of up to
25,000 fps (Figure 4). This equates to an image
capture data rate of approximately 307Gbit/s
– faster than can be transferred from camera In the USA: +1 631 761-5000
Outside the USA: +49.671.601.387
to host computer using any popular commer-
w w w. s c h n e i d e r o p t i c s . c o m
cially-available camera to computer interface.
For this reason, the camera can be equipped
Januar y 2018 27
1/4/18 8:00 AM
 product focus on High-speed imaging

with up to 288GBytes of memory so that at Responding to light

speeds of 10,000 fps, an approximately 20s
image sequence can be captured. At data
rates of 25Gpixels/s, over 7.6s of recording
time can be achieved. Once captured, this
data can be saved in the camera’s 2Terrabyte
CineMag IV non-volatile memory and/or
downloaded to a host computer over a 10Gbit
Ethernet interface.
Similarly, the Fastcam SA-Z from Photron
(San Diego, CA, USA; www.photron.com)
also employs a proprietary CMOS imager
with 1024 x 1024 pixels capable of capturing
12-bit image data. Running at 20,000 fps,
this equates to a data capture rate of approxi-
mately 250Gbit/s. This data is then captured
in the camera’s 128GBytes internal memory
that can be transferred to an optional FAST-
Drive 2TByte removable SSD drive or down-
loaded to a host computer over a dual Gigabit
Ethernet interface. Other companies, such
as iX Cameras (Woburn, MA, USA; www.ix-
cameras.com) also produce such high-speed
cameras, all of which can run at variable
frame rates based on the ROI chosen.
Companies mentioned
Andor Technology
Belfast, Ireland
www.andor.com
Automated Vision Systems
San Jose, CA, USA
www.autovis.com
Cordin
Salt Lake City, UT, USA
www.cordin.com
Emergent Vision
Technologies (EVT)
Maple Ridge, BC, Canada
www.emergentvisiontec.com
Fastec Imaging
San Diego, CA, USA
www.fastecimaging.com
FLIR Integrated Imaging
Solutions
Richmond, BC, Canada
www.flir.com/mv
iX Cameras
Woburn, MA, USA
www.ix-cameras.com
For more information about high-speed camera companies and products, visit Vision Systems
Design’s Buyer’s Guide buyersguide.vision-systems.com
When specifying which model to choose,
systems integrators must, according to
Chris Robinson, Director of Technol-
ogy with iX Cameras, be aware of more
than just frame rates, shutter speeds,
camera-to-computer interfaces and on-
board image memory capability. One of
these is sensitivity or how well the camera
responds to light (see sidebar “Understanding
ISO in a digital world,” p.26 this issue).
For its part, Vision Research specifies this
ISO using the ISO 12232 SAT method, for
both tungsten and daylight illumination as ISO
100,000T, 32,000D when the camera is oper-
ated in monochrome mode. “iX Cameras spec-
ify this sensitivity as 40,000 for its i-SPEED 726
since the “D” is optional, but the “T” is not and
because iX Cameras are tested in daylight, not
quoting “D” after the ISO reading is correct,
but sub-optimal,” says Robinson.
Because camera shutter speeds are often of
the order of microseconds in high-speed imag-
ing applications, ensuring the correct amount of
lighting is present is important. To increase the
Mikrotron
Unterschleissheim, Germany
www.mikrotron.de
Odos Imaging
Edinburgh, Scotland
www.odos-imaging.com
Optronis
Kehl, Germany
www.optronis.com
Photron
San Diego, CA, USA
www.photron.com
QImaging
Surrey, BC, Canada
www.qimaging.com
Vision Research
Wayne, NJ, USA
www.phantomhighspeed.com
Ximea
Münster, Germany
www.ximea.com
Figure 5: Cordin’s rotating mirror-based
camera allows twenty, forty or seventy-eight
2Mpixel CCD images to be captured at frame
rates of 4 million fps.
amount of light, many applications use strobed
LED illumination. However, when deploying
such lighting, systems integrators must ensure
that both camera, lighting and triggering are
tightly coupled to reduce system latency.
To study ballistics, for example, Dr. Chris
Yates and his colleagues at Odos Imaging
(Edinburgh, Scotland; www.odos-imaging.
com) developed a system based on the com-
pany’s SE-1000 camera. With an input latency
of less than 1μs, equivalent to a bullet travel-
ing less than 1mm across the camera’s FOV,
the time between trigger and exposure is short
enough such that the bullet will not enter the
FOV of the camera until after recording is ini-
tiated (see “High-speed vision system targets
ballistics analysis,” Vision Systems Design,
December 2015; http://bit.ly/VSD-BALS).
Interestingly, while the origins of digi-
tal high-speed imaging stem from its film-
based origins, many historically film-based
standards such as the ISO sensitivity stan-
dard have migrated (albeit somewhat unsuc-
cessfully) to the digital domain. Likewise,
some of the concepts such as film-based rotat-
ing mirror cameras have been adapted using
solid-sate imagers, an example of which is the
Model 560 high-speed rotating mirror camera
from Cordin (Salt Lake City, UT, USA; www.
cordin.com) that allows twenty, forty or sev-
enty-eight 2M pixel CCD images to be cap-
tured at frame rates of 4M fps (Figure 5).
In the future, while faster camera-to-com-
puter interfaces will lower the cost of high-
speed imaging for many applications, the use
of cameras with on-board memory and novel
opto-mechanical architectures will remain for
those requiring even faster image capture.

28 Januar y 2018 VISION SYSTEMS DESIGN www.vision-s y stems .com

1801vsd_28 28 1/4/18 8:00 AM

 » E-mail your product announcements, with photo if available, to vsdproducts@pennwell.com | Compiled by James Carroll

Vision + Automation

PRODUCTS
SDK supports GigE Vision
and USB3 Vision cameras 3D cameras feature
Medley is a new SDK that is designed for GigE and
USB3 Vision cameras to enable users to control all the
IP65/67 housing
X30 FA and X36 FA cameras are
functions of ISG cameras with low CPU overhead and
specifically designed for use in
memory requirements. Features include a GUI with
harsh ambient conditions and fea-
viewer and full camera control, sample code for vari-
ture IP65/67 protection class, a GigE
ous API usage models, and barcode decoding.
switch and two GigE uEye FA cam-
Imaging Solutions Group
eras that can be mounted at differ-
Fairport, NY, USA
ent distances. Each equipped with
www.isgcameras.com
a 1.3 MPixel CMOS image sensor,
the 3D cameras feature working dis-
tances from 0.5 to 5 meters, a 100-
Sony releases CMOS cameras watt projector unit on which the
A new series of SXGA cameras enable users to move from CCD to CMOS image sen- two GigE cameras can be mounted
sors. The first camera available in the series is the XCG-CG160, which is based on at different distances, and a soft-
the 1/3” IMX273 global shutter sensor. The IMX273 is a 1.6 MPixel ware development kit for configura-
sensor that can achieve 75 fps via GigE interface. The tion. The X30 is designed for moving
C-Mount camera’s features include defect-pixel correction, objects, while the X36—which fea-
shading correction with peak and average detection and tures the FlexView 2 projector—is
area gain to auto adjust for the target object. designed for stationary objects.
Sony Image Sensing Solutions IDS Imaging Development Systems
The Heights, Brooklands, Surrey, UK Obersulm, Germany
www.image-sensing-solutions.eu www.ids-imaging.com

VIS-NIR coated achromatic lens line expanded

Optimized to correct on-axis spherical and chromatic aberrations, the TECHSPEC
VIS-NIR coated achromatic lens line has been expanded with nine new models.
The lenses are corrected for on-axis spherical and chromatic aberrations. Addi-
tionally, the lenses’ VIS-NIR coating features a transmission range from 400 to
1000nm. Nine models with diameters of 12.7mm, 18mm, 30mm, and 75mm,
with ground or Inked Edge models, are now available.
Edmund Optics
Barrington, NJ, USA
www.edmundoptics.com

www.vision-s y stems .com VISION SYSTEMS DESIGN Januar y 2018 29

1801vsd_29 29 1/4/18 8:00 AM

 + Automation Products
Vision ■

First CMOS image sensor with
Nyxel NIR technology introduced
The OS05A20 CMOS image sensor, the first
sensor to implement Nyxel technology from
OmniVision Technologies, Inc., leverages novel
silicon semiconductor architectures and pro-
cesses that address the inherent challenges in
NIR detection in image sensors. Nyxel com-
bines thick-silicon pixel architectures with care-
ful management of wafer surface texture to
improve QE, along with extended deep trench
isolation (DTI) to help retain (modular transfer
function) MTF without affecting the sensor’s
dark current. The sensor itself is a 5 MPixel
color CMOS image sensor with NIR sensitivi-
ties exceeding 850nm and a capture rate of 60
fps with 2688 x 1944-pixel images.
OmniVision Technologies, Inc.
Santa Clara, CA, USA
www.ovt.com
Scalable line of CMOS image
sensors introduced
Designed for deployment in automotive appli-
cations, the new Hayabusa platform of CMOS
image sensors feature a backside-illuminated
3 µm pixel design that delivers a charge capac-
ity of 100,000 electrons. Other key features
include simultaneous on-chip high dynamic
range (HDR) with LED flicker mitigation (LFM),
plus real-time functional safety and automo-
tive grade qualification. The first product in
the line is the AR0233, which is a 2.6 MPixel
CMOS image sensor capable of running at
60 fps and featuring multi-exposure mode
for >140 dB high dynamic range, full resolu-
tion LED flicker mitigation with 120 dB high
dynamic range, and >95 dB dynamic range
from one exposure.
ON Semiconductor
Phoenix, AZ, USA
www.onsemi.com
Multi-camera solution for
NVIDIA Jetson introduced
Designed for use on the NVIDIA Jetson TX1/
TX2 development kit, the e-CAM30_HEX-
CUTX2 is a multiple-camera solution that tar-
gets applications requiring multiple full HD
cameras. NVIDIA’s TX1 and TX2 can support
up to six 2-lane MIPI CSI-2 cameras simulta-
neously, and as a result, the e-CAM30_HEX-
CUTX2 consists of six e-CAM30_CUMI0330_

Pro duc t S howc as e

A D V E R T I S E M E N T

SUBSCRIBE NOW AT PHORTETM CoaXPress®

Quad Fiber Extenders
WWW.VSD-SUBSCRIBE.COM
• Fully supports CoaXPress
standard & PoCXP
• Supports high-speed bit
rates: 1.25G-6.25Gb/s
• Plug-and-play: no configura-
tion required
• Extends link distance up to
80 km
• Optional AUX channels
• Standard DIN 1.0/2.3
connectors
• -40~80ºC temp OK
• Low 5W pwr consumption

www.phrontier-tech.com

30 Januar y 2018 VISION SYSTEMS DESIGN www.vision-s y stems .com

1801vsd_30 30 1/4/18 8:00 AM

 + Automation Products
Vision ■ Line Scan Cameras
Monochrome or color from 512 to 8160 pixels.
Advantages: Also available:
MOD cameras. These board-level cameras • USB 3.0 SuperSpeed
• Ruggedized connector
feature the 3.4 MPixel AR0330 color CMOS
• No external power supply
image sensor from ON Semiconductor and SK 9170: Gray Scale Line Signal – 1
255

integrated advance image signal processor, ZOOM

M12 lens, and are connected to the e-CAM-

0

HEX_TX2ADAP adapter board using custom- 0

ized Micro-Coaxial cables. From there, the

color monochrome
ome
adapter board interfaces with the J22 con- 1.0

nector on the Jetson TX1/TX2. the 6.4 MPixel IMX178 sensor with 2.4 µm Spectral
tra
0.0
range
ge
400 600 800 1000
100

e-con Systems pixel size and frame rates of 59 fps via USB
Tamil Nadu, India 3.0 and 16 fps via GigE. Other cameras fea-
Universal Color Line Scan Camera
www.e-consystems.com ture the 12.2 MPixel IMX226 sensor with a Scanner
nne Systems
Scan with integrated Bright-field
1.85 µm pixel size and frame rates of 31 fps via or Dark-field
Illumination
Line scan polarization USB 3.0 and 8 fps via GigE. The sensors feature
camera introduced low dark noise of three electrons, combined
A 2017 Innovators Awards Platinum-level with a quantum efficiency greater than 80%.
honoree, the Piranha4 line scan polariza- Basler
Bright-field
tion camera is now in full production. Polar- Ahrensburg, Germany Illumination

ization imaging, according to the company, www.baslerweb.com

10GigE camera series introduced

The first model in the QX series of 10GigE cam-
eras feature the 12 MPixel CMV12000 CMOS Corrosion Inspector Objective analysis
image sensor from ams (CMOSIS), which is a of corrosion
phenomena
global shutter sensor with a 5.5 µm pixel size that
can achieve 335 fps in burst mode and 92 fps
in regular mode via the 10GigE interface. Avail-

Sector 1 Sector 4

significantly enhances detection capability in

many demanding applications. The Piranha4-
2k polarization camera uses Teledyne DALSA’s
Repre-
quadlinear 2048 x 4 CMOS image sensor with sentative
nanowire micro-polarizer filters and captures Test plate Sector 2 Sector 3

multiple native polarization state data without

any interpolation. With a maximum line rate
of 70 kHz, the camera outputs independent
images of 0°(s), 90°(p), and 135° polarization
states as well as an unfiltered channel. Addi-
tionally, the camera is available in Camera Link
Base, Medium, Full, or Deca formats and is 8,
SAN FRANCISCO, CALIFORNIA
JAN 30 - FEB 1, 2018
Visit us in North Hall, Booth 4962
able in both color and monochrome models, the
camera also has an internal image memory of 2
Laser Line, Micro Focus,
GB, and up to 169 images can be buffered at full
Laser Pattern Generators
resolution. At maximum speed, this corresponds Wavelengths 405 – 2050 nm

10, or 12-bit depth selectable. to a recording time of 0.5 seconds.

Teledyne DALSA Baumer
Waterloo, ON, Canada Radeberg, Germany
www.teledynedalsa.com www.baumer.com

STARVIS sensors featured Entry-level high-speed camera

in eight new cameras reaches 500,000 fps
New ace U cameras feature back-illuminated, The tablet-controlled i-SPEED 713 is designed
rolling shutter CMOS image sensors from for applications including drop testing, airbag
Sony’s STARVIS line. Based on two sensors testing and manufacturing that demand the info@SukHamburg.de www.SuKHamburg.com

from the STARVIS line, some cameras feature highest resolution but not necessarily the fast- Schäfter+Kirchhoff develop and manufacture la-
ser sources, line scan camera systems and fiber
www.vision-s y stems .com VISION SYSTEMS DESIGN
optic products for worldwide distribution and use.

1801vsd_31 31 1/4/18 8:00 AM

 + Automation Products
Vision ■

est speed available. The entry-level i-SPEED 713 features a custom

Sales Offices
2048 x 1538 global shutter CMOS image sensor that reaches a Main Office Product Showcase
61 Spit Brook Road, Suite 401 Advertising & Reprint Sales
frame rate of 500,000 fps (4,260 fps at full resolution), combin- Nashua, NH 03060 Judy Leger
ing for 13 GP/s throughput. The camera also features iX Cameras’ (603) 891-0123 (603) 891-9113
FAX: (603) 888-4659 FAX: (603) 888-4659
i-SPEED Software Suite 2.0 PC software for control, analysis, edit- E-mail: judyl@pennwell.com
Publisher
ing, and playback. The high-speed camera is available in an instru- Alan Bergstein INTERNATIONAL SALES CONTACTS
mentation version or an optional (603) 891-9447
Germany, Austria,
FAX: (603) 888-4659
rugged, high-G rated version Northern Switzerland,
E-mail: alanb@pennwell.com
Eastern Europe
designed to resist shock. Executive Assistant Holger Gerisch
iX Cameras Julia Campbell +49 (0) 8847-6986656
(603) 891-9174 FAX: +49 (0) 8801-9153792
Rochford, Essex, UK FAX: (603) 888-4659 E-mail: holgerg@pennwell.com
www.ix-cameras.com E-mail: juliac@pennwell.com
Hong Kong, China
Campaign Manager Adonis Mak
Tom Markley
3D line scan camera features wide field of view (603) 891-9307
852-2-838-6298
FAX: 852-2-838-2766
Featuring a 1400mm (55in.) field of view, the 3DPIXA dual 200 µm HR 3D FAX: (603) 888-4659 E-mail: adonism@actintl.com.hk
line scan camera offers an optical resolution of 200 µm/pixel. Designed for E-mail: thomasm@Pennwell.com
Japan
machine vision inspection, including complete scans Ad Services Manager Masaki Mori
Alison Boyer Murray 81-3-3219-3561
of large, complex and irregularly-shaped (918) 832-9369 FAX: 81-3-5645-1272
objects or textures, the FAX: (918) 831-9415 E-mail: masaki.mori@ex-press.jp
E-mail: alisonb@pennwell.com
stereo camera features Israel
List Rental Dan Aronovic (Tel Aviv)
two RGB tri-linear CCD Kelli Berry 972-9-899-5813
line scan image sensors, (918) 831-9782 E-mail: aronovic@actcom.co.il
FAX: (918) 831-9758
which feature a 10 µm E-mail: kellib@pennwell.com Should you need assistance with
pixel size and line scan frequencies of up to 30 kHz North American Advertising creating your ad please contact:
at full resolution. The 3DPIXA dual 200 µm HR features a Camera Link & Sponsorship Sales MARKETING SOLUTIONS
Judy Leger Vice President
base/medium interface and can simultaneously acquire 2D color images
(603) 891-9113 Paul Andrews
along with either a height map or 3D point cloud. FAX: (603) 888-4659 (240) 595-2352
Chromasens E-mail: judyl@pennwell.com Email: pandrews@pennwell.com

Konstanz, Germany
www.chromasens.de/en

Machine vision camera line launched Advertisers Index

by Industrial Vision Systems Advertiser / Page no.
Available in five color and five monochrome models, IVS-NCGi cam-
Alysium-Tech GmbH......................................................................4
eras are equipped with either a GigE or USB interface. The cam-
Automated Imaging Association .................................................14
eras feature sensors ranging from 1/3” to 1/1.2” in size and 2.3
Computar Optics by CBC Americas...............................................8
MPixel to 5 MPixel in resolution, with frame
Edmund Optics ............................................................................. 2
rates reaching 181 fps in the USB
Intercon 1 ................................................................................... 17
models and 60 fps in the
Jargy Co., Ltd............................................................................ CV3
GigE models. IVS-NCGi
Landesmesse Stuttgart GmbH .................................................... 11
cameras come ready
Matrox Imaging ....................................................................... CV4
to be mounted with
Mikrotron GmbH ........................................................................24
standard LED lighting
Osela Inc. ......................................................................................4
options, plus a range of Photonfocus AG .......................................................................... 13
field-changeable C-Mount lenses and industrial auto- Phrontier Technologies................................................................30
focus lens options. Additionally, the cameras come with IVS soft- Schäfter+Kirchhoff GmbH........................................................... 31
ware platforms, which enable set up and integration for inspection in Schneider Optics ......................................................................... 27
machine vision applications. Sill Optics GmbH & Co KG...........................................................20
Industrial Vision Systems Vieworks .................................................................................. CV2
Oxfordshire, UK This ad index is published as a service. The publisher does not assume any
liability for errors or omissions.
www.industrialvision.co.uk

32 Januar y 2018 VISION SYSTEMS DESIGN www.vision-s y stems .com

1801vsd_32 32 1/4/18 8:00 AM

1801vsd_C3 3 1/4/18 8:07 AM
 Why walk
when you can fly?
Meet MIL CoPilot, the interactive environment for Matrox Imaging
Library (MIL). Machine vision application developers can plan their
course by readily and efficiently experimenting and prototyping
with MIL—all without writing a single line of program code. With
the trajectory set, MIL CoPilot accelerates the journey towards
application deployment with a code generator that produces clear,
functional program code in C++, C#, CPython, and Visual Basic®.
Developers using MIL get projects off the ground quicker and
easier than ever before with MIL CoPilot and ready access to
Available as part of Matrox Vision Academy online training.
MIL

Get there faster with a CoPilot

www.matrox.com/imaging/co-pilot/vsd

1801vsd_C4 4 1/4/18 8:07 AM