- » View All
Color Vision Tackles Line-Scan Applications
by Winn Hardin, Contributing Editor - AIA Posted 12/11/2015Despite the fact that people see the world in color, machine vision system designers traditionally eschew color solutions for simpler, more economical monochrome solutions. Color line-scan cameras typically cost more than their monochrome counterparts, especially when trying to maintain the same spatial resolution. They also produce at least three times more data (red, green and blue) than monochrome solutions, which adds more complexity and cost to the processing elements, as well as requires additional image-processing knowledge and engineering time.
“When key material properties are encoded in color, then you need a color solution,” says Ben Dawson, director of strategic development at Teledyne DALSA (Waterloo, Ontario, Canada). “Because color generates three times the data, any improvement in processing speed is going to open up new applications in color.”
Teledyne DALSA is currently doing an application that inspects 20 bottles per second, including the label. “Just four years ago before the most recent Intel processor cycle, we either couldn’t have done this application, or we couldn’t have done it at that speed,” Dawson says. “Today, using our GEVA 3000 high-performance vision system with real-time OS, extra hardware, and high-speed memory transfers, it’s not a problem.”
Label inspection is one of three high-volume color applications served by Matrox Imaging (Dorval, Quebec, Canada). Applications that benefit from color imaging include packaging inspection for label aesthetic, produce inspection for grading visual appearance, and assembly verification of wire harnesses, for example, says Pierantonio Boriero, product line manager for Matrox Imaging.
“While the availability of more computational power at a lower cost has increased the practicality of color imaging, the main obstacle to adoption remains the technicians’ understanding of, and thus comfort with, color imaging technology — that is, knowing what it can and cannot do in relation to human perception, which color is ultimately all about,” Boriero says.
Perception is at the heart of any color vision solution. In particular, will the system try to mimic human color perception, judge one color against another in relative terms, or make absolute colorimetric measurements?
“There are two types of color calibration,” explains Teledyne DALSA’s Dawson. “One is close enough; the other is absolute. With absolute, the customer says they need the system to determine a color with these CIE color coordinates within a given tolerance.”
For colorimetric applications, Dawson suggests placing a ceramic color reference inside the camera’s field of view for regular calibration. The lights, usually LED, also require careful monitoring, and both should be in temperature-controlled enclosures to ensure temperature variations don’t affect hardware performance. Finally, the angle of presentation between the light, object under test, and camera also are important.
“If you’re looking at vegetables that present differently to the camera, you have a problem. If it’s a manufactured item and you control the presentation, presentation is not an issue,” Dawson says. For precise color measurements, most customers will use a backup offline color measurement system to periodically verify the machine vision system’s findings.
Colorimetric applications typically require a stable white light with flat spectral output across the visible portion of the spectrum. According to Mark Kolvites, technical sales manager at Metaphase Technologies, Inc. (Bristol, Pennsylvania), most clients seeking a light source for colorimetric applications prefer the company’s cool-light diffuse dome LED lights as a light source.
“Our dome lights provide very even illumination with sufficient [color rendering index] to accurately judge colors,” says Kolvites. “This is particularly useful for vegetable inspection lines that cover a mix of vegetables instead of just one type. It’s also great for color web inspections that may include up to 13 different color separations on a single web. A uniform white light source is critical for those applications.”
Kolvites’ associate, Kevin High, adds that for absolute measurement, the application needs to not only use high-quality cool white LEDs but also choose the chips from the same bin to ensure uniformity in both color temperature and flux.
Even when the system measures relative instead of absolute color, calibration of camera and light source is still important. According to Matrox’s Boriero, the company’s latest MIL software development kit (SDK) includes a straightforward and robust method for automating color calibration to ensure consistent performance – not just within a single system but across multiple systems over time. “MIL works in LAB color space because it better represents human perception when it comes to color,” adds Boriero.
As color machine vision applications continue to grow, vision suppliers are expanding the capabilities of their systems to meet demand. With automated calibration routines or self-aligning color cameras such as Teledyne DALSA’s Piranha XL 16k that not only help align the camera to the color space but also orthogonal to the physical direction of the underlying conveyor, it’s clear that vision suppliers are poised to match the momentum of color applications.
There are currently no comments for this article.
Leave a Comment:
All fields are required, but only your name and comment will be visible (email addresses are kept confidential). Comments are moderated and will not appear immediately. Please no link dropping, no keywords or domains as names; do not spam, and please do not advertise.