Vision education: It's almost free
Assemble a machine-vision learning plan, then tap vendors for free information., Contributing Technical Editor
Jon Titus -- Test & Measurement World, 11/1/2003
When you need to learn more about machine-vision hardware and software, don't count on local colleges and universities to offer a degree program or even a semester course. Few, if any, such programs exist. Instead, you can tap into a wealth of information provided by vision-equipment manufacturers and vision-software suppliers. Although you may have to pay for some seminars and short courses, a lot of the material is available at no charge. (See, "Machine-vision resources.")
The breadth of information from vendors can overwhelm a newcomer, so to start, divide machine vision into two subject areas—image acquisition (hardware) and image processing (software). The image-acquisition portion of an education program encompasses topics such as lighting, cameras, lenses, and frame grabbers. The image-processing part of a curriculum will help you learn how to extract and use information from an image.
 |
|
State-of-the-art vision equipment and computers let students experiment with concepts and techniques describes in lectures. Students can help each other and work with instructors on real-world problems. Courtesy of DVT. |
Although engineers who develop vision systems require knowledge in both areas, mastering the fundamentals of image acquisition takes the most effort. Fully understanding the interactions between lights, lenses, cameras, and frame grabbers requires hands-on experience, but you can learn the basics by reading application notes, data sheets, and tutorials about lighting techniques, lens configurations, and camera types.
Most people involved with machine-vision applications agree: Image-acquisition techniques take the most effort to get right. Robert Rochon, an applications manager at Matrox Imaging, says, "When you acquire good quality video images, it simplifies any subsequent image processing. Capturing a good image is a real milestone, and it can typically take about half of the development effort to get there."
One key to obtaining a good image is selecting the right combination of equipment. Suppliers of lenses, frame grabbers, and cameras often provide application information that you should study carefully. In addition to describing specific products, data sheets and catalogs often contain application information, too. Some suppliers also provide useful selection guides. For example, the Graftek Imaging Web site includes an interactive lens-selector worksheet that can help you find a lens based on information you supply about an application's camera type, distance to subject, depth of field, and so on.
Frame-grabber and camera suppliers have similar selection guides to help you select compatible products. Vendors' Web sites also list application notes that can help you understand how cameras work and determine what types of cameras suit particular applications. Matrox Imaging, for example, offers two camera-interfacing guides on its Web site.
The I/O operations that control cameras and lights and communicate with external devices require study, too. Dr. Cliff Fitzgerald, the senior manager of worldwide educational services at Cognex, says newcomers may not have experience connecting the various elements of a vision system. Members of his company's technical support staff often answer questions about connecting vision systems to discrete I/O devices, programmable logic controllers (PLCs), and Ethernet networks. Companies that make I/O boards, Ethernet cards, and vision-system hardware provide application information about interfacing techniques.
Extract some dataAfter you feel you have a good grounding in the hardware side of machine vision, you can move on to learn about image processing. Here, your education should concentrate on how to extract useful information from images. In general, image processing involves using well-defined algorithms, also called tools, that measure, count, and compare features in an image. Software also can obtain information from bar codes or machine-readable characters.
 |
|
Seminars give participants an opportunity to talk with fellow vision-system users and exchange ideas in a classroom setting. Instructors can schedule time to help individuals with specific machine-vision applications. Courtesy of Cognex. |
Before you can effectively use vision tools, you must define what you want to achieve as a result of inspecting an item. As you learn about software, think about how to approach simple tasks, such as measuring an angle, counting items, matching simple shapes, and so on. If you first try to solve a complex problem, such as identifying solder defects on a PCB, you'll get lost before you understand what vision tools can do.
Although an introduction to vision software starts with reading application notes and watching demonstrations, there's no substitute for hands-on experience. When possible, work with demonstration software and manipulate test images—your own or those supplied with the demo software. (Of course, you need good images if you want to obtain useful results.)
If your plans call for extensive software development, and demo software comes with sample applications, pick the applications apart to learn how the developer used specific tools to solve a problem. The National Instruments' Web site provides a development library of machine-vision software written in C, C++, and LabView. You can download the code and read short application notes that explain how to solve problems such as gauging, edge detection, or optical character recognition. You'll find that other vendors' Web sites also include application notes that describe how to apply tools to visions problems in various industries.
Catalogs, courses, and conferencesIn addition to Web-based application notes, videos, and selection guides, many companies offer CD-ROMs and paper catalogs. The catalog from Edmund Industrial Optics, for example, includes more than 30 short tutorials on vision and optics topics. Cognex distributes a VisionGuide CD, and Keyence offers a "Machine Vision Encyclopedia" on CD-ROM. Even CD-ROMs and catalogs that contain mainly product information can help you understand the characteristics and specifications of vision-system components. Product photos, for example, may suggest mounting arrangements for cameras and lights.
If you seek education that allows for interaction with a teacher, investigate vendors' courses that offer lectures and lab exercises that use real equipment. There's no substitute for first-hand experience. If you attend a course, take along examples or images of any products you plan to inspect. Instructors will talk with participants about their vision problems and how to solve them. Lunch breaks also give attendees time to discuss their needs with fellow engineers.
DVT's Henry stresses that all his company's courses are free, and anyone who wishes can participate, whether or not they have a technical background. Courses range from beginner-level material to advanced training, and students get to use the latest lighting equipment, lenses, and cameras from many vendors.
National Instruments offers a two-day course that travels from city to city in the US. Invited vendors demonstrate equipment such as lights and cameras so students see how real equipment operates together. Jason Mulliner, NI's vision marketing manager, says the first hour or two covers the art of acquiring images and the rest of the course covers image-processing techniques. The course requires some experience with the company's LabView software. Mulliner notes that NI will soon offer machine-vision courses that require little or no exposure to LabView.
Cognex also offers hands-on courses, from a one-day introductory course to five-day courses that cover the operation and applications of specific Cognex products. Of all the companies that offer vision training, only Cognex offers continuing-education units (CEUs). Course prerequisites range from no experience to some experience with machine vision. If you plan to program with C, C++, or Visual Basic, consider the three-day courses offered by Matrox Imaging that explain the use of the Matrox Imaging Library.
You also can get first-hand experience at shows and conferences sponsored by the Automated Imaging Association. Conference sessions give attendees opportunities to hear how others have solved machine-vision problems and to talk with vision experts. At the exhibition booths, vendors' representatives demonstrate new products and answer questions and talk with attendees about inspection, cameras, and lighting.
There's plenty of useful information to keep you busy learning about machine vision. As always, though, the quality of the education depends on how much effort you put into it.
| Where to find information | ||
| The following companies were mentioned in this article: | ||
| Automated Imaging Association Ann Arbor, MI www.machinevisiononline.org |
Cognex Natick, MA www.cognex.com |
DVT Duluth, GA www.dvtsensors.com |
| Edmund Industrial Optics Barrington, NJ www.edmundoptics.com |
Graftek Imaging Austin, TX www.graftek.com |
Keyence Woodcliff Lake, NJ www.keyence.com |
| Matrox Imaging Dorval, QC, Canada www.matrox.com |
National Instruments Austin, TX www.ni.com |
|
| For more information | ||
| Other vendors of machine-vision hardware and software also offer educational material on their Web sites. You can find these companies in the T&MW Buyer's Guide. Look under the Inspection Equipment and Software headings: www.tmworld.com/bg. | ||
| In addition, the T&MW Web site also includes an archive of articles about inspection techniques and machine-vision components: www.tmworld.com/ins. | ||
