VISION 2010: a study in variations of machine vision

As machine vision technologies and products become more established across multiple industries, tradeshows such as the forthcoming VISION 2010 to be held November 9-11 at the New Stuttgart Trade Fair Centre in Stuttgart, Germany, will reflect these trends.

Indeed, during VISION 2010, a panel discussion entitled: “Green Vision – Driving Factor for a Green Future” will focus on how machine vision can be used in systems to protect the environment, conserve resources, increase energy efficiency, and develop more environmentally friendly products.

In addition to highlighting innovations in industrial camera and system design, the show will also include a demonstration of autonomous robot footballers, an application park highlighting the role machine vision plays in testing and production processes, an area demonstrating international machine vision standards, joint booths for startup companies, and a series of seminars for those new to machine vision.

According to the organizers, Messe Stuttgart, attendance is already on track to exceed last year, both in terms of exhibitors (now over 300) and attendees. Those who wish to see the many sides of machine vision would do well not to miss the event.

Hyperspectral imaging heads commercial

Multispectral imaging enables several discrete images in the visible and IR bands of the spectrum to be captured and processed. To capture continuous spectral bands from the ultraviolet to the far infrared, hyperspectral imaging is a powerful if often expensive imaging tool.

Hyperspectral remote-sensing applications have flourished for several decades. Now, low-cost imaging spectrometers are being introduced that allow innovative approaches to applications such as medical diagnostics, metallurgy, sorting materials, food processing, and microscopy.

We recently published an article by Rand Swanson at Resonon describing a compact hyperspectral imaging system that can be flown in a Cessna aircraft to monitor the spread of leafy spurge, an invasive weed that reduces grazing forage for livestock.



We’ve also reported on the use of hyperspectral imaging to detect the food pathogen Campylobacter and to sort walnuts.

A hyperspectral imaging microscopy system also allows detailed examination of LED structures in the visible and near-IR.



You can find more examples by searching our website. I expect to see numerous such articles in the future. For example, we’ll be describing a hyperspectral blueberry sorting system from EVK in Austria in our September issue.

Big Pharma needs machine vision

Discussions on the Vision System Design Group on LinkedIn have recently reflected the growing interest in using machine vision to inspect pharmaceutical products. We have published a series of technical article that might be of interest.

One article about a pharmaceutical packing system that uses IR and visible sensors describes how American SensoRx developed a system that inspects tablets, capsules, caplets, and gels at very high speeds before they are packed up and shipped to distributors.


Another describes how the German company Boehringer uses FireWire cameras to inspect capsules used in inhaled medications for respiratory disease.

And yet another describes how Pfizer added an x-ray system to its visible light inspection system to check tablets in blister packs.

Robots, with vision, at your service

A recent article and video in the New York Times describes Bandit, a robot built by researchers at the University of Southern California, which interacts with autistic children. Three-foot-tall Bandit can maintain “eye” contact with an autistic child and, sometimes, use playful or sympathetic actions to overcome withdrawn behavior.

Another robot, named RUBI—Robot Using Bayesian Inference—at the University of California, San Diego, images children’s faces, recognizes basic emotions from facial muscle movement, and responds with verbal and physical gestures of encouragement.

These service robots are part of a rapidly growing wave of robotic human helpers. In the classroom they may supplement the work of human teachers, during surgery they may perform delicate procedures, and on the battlefield they may help disarm a roadside bomb, as described in our June 2010 cover story.

The technological differences between these service robots--with their vision and image processing functions--and robots used in industrial applications can be small. For example, a recent article on our website describes the work of researchers at the Technical University of Munich who are imaging non-verbal communications such as gestures and facial expressions as a method of interacting with robots. To date, they have demonstrated that their work can help those that require assisted living and workers in automated production plants, where background noise may make speech recognition difficult.

Recently, European researchers have built a robot for 'on-demand' rubbish collection – just make a call and it will soon arrive at your door. It's ideal for collecting waste in the narrow streets of many historical towns.



About the size of a person, it can navigate the narrowest of alleys, stop outside your door and take your rubbish away. And the best bit is this: You don't have to remember when to put your bin out, but simply make a telephone call. Soon the robot is waiting outside your door, ready to receive your rubbish.

July VSD online--Hyperspectral imaging, minature autofocus lenses, 3-D vision

To capture continuous spectral bands from the UV to the far IR, hyperspectral imaging has become a powerful imaging tool. In our July issue, Rand Swanson at Resonon describes a compact hyperspectral imaging system that has been flown in a Cessna aircraft to monitor the spread of leafy spurge, an invasive weed that reduces grazing forage for livestock.

In our Product Focus article, editor Andy Wilson describes recent developments in miniaturized autofocus lenses. Whether based on electro-optical, electromechanical, thermo-optical, or acousto-mechanical techniques, these tunable optics will find cutting-edge applications in smart machine-vision systems, endoscopy systems, and mobile phones.

Our cover story shows how an optical tester based on an off-the-shelf camera system can be used to calibrate centering errors of lenses to ensure the imaging quality of an optical assembly or subassembly.

3-D vision remains one of the most alluring areas for innovation in machine vision development. While dual-camera and time-of-flight sensors are becoming increasingly important, other options such as the one described in an article about ISee3D now allow stereo images to be captured from a single camera/lens combination.

We also have articles on: inspection of wood surfaces for defects by researchers at AIDO in Spain; an algorithm that uses partial deriatives to improve edge detection; and an FFT processor that performs phase correlation.

Blogging about machine vision – interested?

Some cynics I know mock the idea of blogging, but I think it’s a good way to explore a subject such as machine vision. And a blogger might even be paid the highest compliment--having your blog blogged about.

A case in point: editor Andy Wilson’s My View video blog on the Vision Systems Design website was recently blogged about by Laura Hoffman, who runs the Microscan blog SolutionConnection, along with colleagues such as John Agapakis.

Numerous other companies in the machine vision industry have blogs or are trying to figure out what they could write that wouldn’t rattle internal corporate feathers but would still be interesting. Like Microscan, Thor Vollset at ScorpionVision’s blog is aiming to keep readers up to date on the company and how customers can use its products.

Then there are system integrators who have occasional blogs on their own sites or on a magazine site. These include David Dechow at Aptura Machine Vision Solutions with his Regarding Machine Vision blog, and Ned Lecky at Lecky Integration and John Nagle at Nagle Research. Also, there are journalists who blog about related topics, such as Frank Tobe at Everything Robotic and Gabriele Jansen at Inspect-online blog.

And of course there is the popular and anonymous B Grey at machinevision4users blog, who presumably hides his or her identity out of concern about industry (or employer?) reaction. The postings vary from technical observations and comparisons to witty digs. But anonymity is both a shield and a crutch. Speaking as a journalist who must live with the consequences of what I write, I think B Grey should stand forth and be counted.

Whether anonymous or very public, all bloggers can attest to the fact that it’s not easy to post frequently and have something interesting or new--or at least amusing--to say. Yet it can be quite rewarding, personally and professionally.

Blogs can become good networking and marketing tools that engage people. And you can re-post blogs to other social media sites such as Linked In, where you will find many relevant groups such as the Vision Systems Design Group, the Machine Vision Group, the Image Processing Group, and the 3D Machine Vision Group. Most of these groups have hundreds or even thousands of members.

If you have a comment on what I’ve written, please post it on this blog.

If you’re reading this and interested in contributing a regular or at least somewhat regular blog to Vision Systems Design, please let me know: cholton@pennwell.com.

June issue: vision-guided robots, software, and Three Mile Island

Our June issue is now available on our website. The articles in it point to some of the many ways in which machine vision is evolving.

I glimpsed this potential in 1982, when I watched the feed from the first remote video camera lowered into a reactor vessel at Three Mile Island, after the nuclear accident had destroyed the reactor core in 1979. It took several years to develop the imaging equipment for this first foray and, in the years that followed, many cameras and robots would gather information about damage and perform cleanup operations in highly radioactive areas of the plant. Here's a picture of Rover, developed with Carnegie Mellon University.


Although robots were not then sophisticated enough to perform major operations—and stereo vision was practically a dream—the future of vision-guided robots was obvious. Some colleagues and I wrote a history of the cleanup, including the robotic and imaging technologies that were used. You can download a PDF of the history published by the Electric Power Research Institute by clicking HERE.


Remotely operated vehicles are now playing an increasing role in other crises. Our cover story in the June issue, for example, shows how 3-D displays can help remote operators in the military safely handle and dispose of explosive devices using robots.

Another article explains how single-sensor image fusion technology could enable simpler and more effective imaging of potential threats in security and defense operations.

Machine vision is not always on the front line of environmental and political challenges, however. Researchers from the University of Ilmenau in Germany are using image processing techniques to evaluate the quality of wheat after it is harvested.

And, as contributing editor Winn Hardin explains, manufacturers are using other machine vision techniques ensure that the steel tubes produced for oil and gas production are of the highest quality.

This broadening range of biomedical, robotics, military, and aerospace applications is leading software vendors to expand the functionality of their products beyond simple measurement functions, as editor Andy Wilson writes in his Product Focus article on machine vision software.

Indeed, new opportunities for machine vision and image processing systems are occurring every year. To take advantage of these developments, however, suppliers of machine vision systems will have to look outside the box of conventional industrial manufacturing and into niche applications that span the gamut from agriculture to space exploration.