The combination of cheaper, larger
sensors - thanks to maturing semiconductor manufacturing processes
and an explosion of digital interfaces - is pushing machine vision
camera suppliers to broaden their product offerings. But is having
more options always a good thing?
According to Michael Cyros, Chief
Commercial Officer of Allied Vision Technologies (AVT), a Stadtroda,
Germany-based provider of high-performance digital cameras for image
processing, diversification among machine
vision cameras is both a benefit and challenge to the growing
AVT’s Bonito 4 MP high-speed camera featuring CMOS sensor.machine
vision market. With a growing, diverse customer base wanting more
camera variations for their own needs, customers will be challenged
to find the right camera, requiring suppliers to step up as
consultants and offer more guidance than ever before.
More Pixels, But Bigger, Too?
The first of several trend drivers is
the need for increased resolution, oftentimes combined with the need
for increased sensitivity, Cyros notes. This is leading to
higher-resolution sensors as well as larger optical formats, reaching
beyond the needs of traditional C-mount optical formats.
Much of this push is coming from
high-end applications - printed circuit board inspection, flat panel
inspection, and so on - mainly so users can inspect more and more
detail in these areas.
“What they have to inspect keeps
getting smaller and smaller, which means more pixels, which means
more camera
link repeater per system,” says Mark Butler, Product Manager
with Teledyne Dalsa, a Waterloo, Canada-based provider of digital
imaging solutions, including cameras and frame grabbers. “It’s
getting to the point where the number of cameras is getting
unwieldy.” As a result, he notes, manufacturers have been trying to
get more pixels into a camera.
The need for larger sensor arrays is
pushing some camera makers toward pixel shrink. “In some cases, the
objects users are looking at are so small, it’s to their benefit
for us to shrink the pixel because that will help to lower their
optics cost,” Butler says. “We’re always trying to manage that
- how large the pixels should be relative to the cost of the optics
and so forth.”
Rick Roszkowski, Senior Director of
Marketing, Vision Systems Business Unit at Cognex Corp.'s West Allis,
Wisconsin facility, specializes in Cognex Machine Vision System In-Sight smart cameras,
including the DataMan AutoID image-based code readers. His customers
want to be able to read barcodes across an entire carton or pallet at
one time. This means cameras with higher resolutions.
“VGA is losing favor to 1.3- and
2-megapixel [MP] cameras, which puts more pressure on the storage if
you’re keeping images, and processing capabilities as well,”
Roszkowski explains. “Processing 120 barcodes at once can take 2.5
seconds, for example. So, in some cases, customers are saving the
images and performing post-processing offline to determine whether a
bad printhead is the cause of a defective barcode. The image-based
reader runs faster than the JAI's SP-20000 is a 20-MP camera built
around the CMOSIS CMV20000 full-frame imager, delivering 30 frames
per second at full resolution while maintaining a 6.4-micron pixel
size.case packer, so that’s not a problem. But for more detailed
image analysis, sometimes it makes sense to run that offline.”
More Pixels, More Bandwidth, Please
“We’re redesigning some of our
camera front ends to deliver both increased resolution without
sacrificing frame rate and speed, such as our new SP-5000 5-MP camera
that can go up to 209 frames per second today and will go to 250
frames per second very soon,” says Steve Kinney, Director of
Technical Pre-Sales at JAI, Inc., a San Jose, California-based
provider of digital CCD/CMOS camera technology.
The advent of CMOS sensors is allowing
camera designers to cost-effectively adopt larger sensors that
operate at higher frame rates. “The growing selection of
high-resolution global shutter CMOS sensors is starting to displace
traditional CCD sensors in certain applications,” says AVT’s
Cyros. “The image quality of these new generations of CMOS sensors
has made remarkable improvements in recent years, which is why we are
finally seeing a big increase in demand for CMOS sensors for higher
performance-type applications, instead of just the low-end
application price points where CMOS traditionally fits.”
The most common way to offer more
resolution while maintaining costs is to leverage improvements in
semiconductor manufacturing and volumes to reduce cost or shrink the
pixel size.
New high-speed, high-performance
line-scan imaging cameras such as JAI’s SW-2000 2k line-scan camera
with 20-micron pixels present another option. “That’s 400 square
microns of active area compared with 50 for a 7-micron pixel,”
Kinney explains. “And that means eight times the sensitivity
improvement or throughput speed, or some combination thereof.”
The higher speed of new digital
interface specifications is helping companies like JAI, Teledyne
DALSA, and AVT to push camera designs into new territory, opening up
new application areas. But it can make it more challenging for
customers to choose the right combination.
“Since Camera Link , there have been
a number of new standards available,” notes Cyros. “Rather than
converging, each interface finds its way into certain application
areas. Today, users are faced The high-resolution Prosilica GT4905
camera from AVT featuring 16 MP Truesense KAI-16050 CCD sensor.with a
range of standards choices, including GigE Vision, USB3 Vision, and
CoaXPress, in addition to Camera
Link Repeater and others. This presents a challenge for a camera
manufacturer that wants to address a wider range of the market.”
Cyros also points to the growing use of
machine vision camera technology
in non-industrial application markets to illustrate his point. “There
are many newer applications such as specializations in medical, ITS
(intelligent transportation systems), sports, and entertainment that
require precise image capture in difficult lighting conditions, often
in harsh environments, that are benefitting from the feature sets
that are relied upon by the traditional machine vision market needs.
“In the end, this wide range of
trends promises to keep things interesting for camera manufacturers
as we all seek to find ways to differentiate ourselves from each
other,” Cyros concludes. “This will likely lead to continued
fragmentation in the camera market, making it difficult for camera
manufacturers to achieve dominant positions. Combined with an
increasingly lower barrier to entry for new camera suppliers, along
with a growing range of interface standards and applications types,
the camera user may be finding it increasingly difficult to make the
right camera choice for their application needs, putting additional
pressure on camera manufacturers to make sure their customers
understand the unique performance of each model and guide the
customer to the right selection.”
Contact us at:
Menzel Vision & Robotics Pvt Ltd
Address: 4, A-Wing, Bezzola Complex,
Sion Trombay Road, Chembur
400071 Mumbai, India
Tel:(+91) 22 67993158
Fax: (+91) 22 67993159
Mobile:+91 9323786005 / 9820143131
E-mail: info@mvrpl.com
