During my writing tenure on Vision Systems Design, I frequently have been asked by readers to assist them in helping to develop machine-vision systems. Whereas many of these questions have been direct and to the point, such as asking for information on standards such as Camera Link or on manufacturers of PCI frame grabbers, others have been less specific. Most of these questions emanate from systems integrators attempting to build state-of-the-art vision or imaging systems.
To help answer such questions, several Web sites dealing with machine-vision and image-processing products and technologies, including that of the Automated Imaging Association (AIA; www.machinevisiononline.org), have special sections to address this need.
When presenting technical questions, readers want straightforward answers. Recently, for example, a reader needed product information on a CMOS imager with a pixel size of 15 µm, a form factor of at least 640 x 480 pixels, and a 10-bit analog-to-digital converter on the chip.
Another, more complex, example comes from Teknor Apex (Pawtucket, RI), a supplier of garden hose. It needed a means to inspect hose for blemished and blistered sections while it was moving on an assembly line at 400 ft/min. At this rate, the problem is not trivial, especially as requirements mandated the inspection of the entire circumference of the hose. As expected, typing "automated garden-hose inspection systems" into the latest Web-based search engine yielded no results.
One possible solution is a system that not only controls large lengths of hose traveling at 400 ft/min for long periods of time, but also inspects and flags faults for later machine processing. As with most complex, high-speed automated production equipment, the cost of the machine-vision system is usually small compared with the cost of handling products being inspected at fast rates. When examined from this aspect, what is really required is a system that can, under computer control, handle the production of hose at fast rates.
Interestingly, many of today's steel-pipe inspection systems are already being used to test, classify, and mark pipe as it is manufactured at speeds of up to 600 ft/min. In many of these systems, operators monitor and control product testing using a keyboard or a touch-sensitive color screen. As sections of the pipe are imaged by ultrasound detectors, a map is produced on the screen in real time, indicating both thickness and flaw exceptions. While such machines detect flaws in steel, they might be adaptable to detect blemished and blistered sections of hose using a machine-vision system linked to a host control computer.
So, costly materials-handling capabilities already developed for one type of inspection system may provide a solution for a manufacturer of similar, yet different, materials. In formulating machine-vision inspection questions, the real problem may be obscured by the end user's desire to automate product inspection. What may be a more important first question for systems integrators may be one that pertains to high-speed materials handling rather than one about how to perform machine vision.
Indeed, what is needed is a cross-fertilization of ideas among vision systems designers, automation equipment manufacturers, and production plant operators. To obtain a complete perspective, systems integrators would have to attend multiple trade shows and receive multiple trade publications. Vision Systems Design attempts to achieve this cross-fertilization by presenting a number of different types of systems built for various applications.
What are also needed, though, are industry groups and associations to join forces to present complete coverage of the systems-integration process. For the vision/imaging industry, this might be accomplished by merging a number of industrial, robotic, and vision trade shows to present attendees with both OEM products and their typical applications.
