Inspection Systems Benefit from High-tech Components
Machine builders wanting to differentiate themselves are always looking for the best technology for the project at hand. Machine vision systems using state of the art hardware and software can help a machine builder gain an edge with regard to inspection systems. In this article, we’ll explore two different solutions for two very different quality control applications.
One company uses CompactPCI (CPCI) hardware to enable paper and metal surface inspection at 6,000 ft/min. Another makes sure wine bottle labels are straight and correct with the help of sophisticated analysis software.
The top 10 metal producers and 15 of the top 20 paper producers rely on German machine builder Parsytec to provide them with surface inspection systems. Parsytec’s “espresso SI” surface inspection machine is capable of handling strips of paper or metal up to 40 ft. wide at speeds up to 6,000 ft/min, and its imaging system needs to keep pace.
For defect recognition in surface inspection applications, cameras need to provide a resolution of 250
“Prior to the use of digital cameras with gigabit Ethernet (GigE) capabilities, inspection systems used standard industrial PCs with framegrabbers and analog cameras. These PCs were housed in large cabinets and used fault-prone hard drives and cooling fans. The introduction of digital cameras to the industry eliminated the need for these unreliable system components. They also freed up a system slot by eliminating the need for framegrabber hardware,” according to Stephen Cunha, vice president of MEN Micro Inc.
Processing power importance
After Parsytec reviewed embedded computing boards available in the small CPCI form factor that could be synced up to form a parallel computing environment, it chose MEN Micro’s F15 CPCI single board computer (SBC) to serve as the central control of the imaging system. These types of boards eliminate the need for hard drives that can fail and for cooling fans typically used for each board—two components that negatively impact system reliability and ease of maintenance, says Cunha.
Jan Erxleben, manager R&D for Parsytec, said, “MEN Micro’s CompactPCI SBCs enabled us to fit 15 boards in one 4U, 19-in. rack vs. mounting 15 industrial PCs in two large cabinets. And the board had the fastest Intel Pentium processors available at the time: 2.0 GHz, when everyone else was shipping only up to 1.8 GHz.
“Processing power is central to our imaging component,” added Erxleben. “We need to provide our customers with the best, most efficient solutions available.”
MEN Micro’s 3U F15 model is part of the company’s CPCI family of embedded computing boards that now incorporate Intel Core 2 Duo processors and Intel 945GM Express chipsets for high-quality graphical computing performance. There is 2 MB of L2 cache integrated in the Core Duo processor, and up to 4 GB of soldered SDRAM DDR2 system memory operating over a 667 MHz bus frequency, says Cunha.
The F15 board offers a variety of input options for added flexibility including two serial advanced technology attachment (SATA) interfaces, video via VA and two serial digital video output (SDVOs), two GigE ports (critical for capturing the number of images required in surface inspection applications) and eight USB 2.0 ports.
Erxleben says that, with CPCI technology, his customers have seen a dramatic increase in system reliability and mean time before failure (MTBF), as well as a reduction in maintenance requirements and overall cost of ownership. If a unit does go down, the defective board also can be replaced much easier, without costly data back-up-and-restore processes associated with traditional industrial PCs.
Erxleben says Parsytec saves its customers several million dollars per year with the quality of data its surface inspection systems can produce. Plans to migrate to the latest generation of MEN Micro Core 2 Duo processing boards, the F18, are underway. On top of the advanced functionality and scalability provided by the F15, the F18 SBC uses Intel’s T7500 Core 2 Duo processor with a frequency of 2.2 GHz for even more processing power to keep pace with evolving inspection requirements.
Bottle label inspection
The Kendall-Jackson winery (Fulton, CA) depends on machine vision to increase efficiency in areas such as bottle inspection. The company needed a solution for checking labels after they had been placed on the bottle. They chose CI Vision’s 360 Full View, a wine bottle inspection system that is flexible enough to inspect labels independently of a bottle’s orientation on a conveyor.
CI Vision (Aurora, IL) uses analysis software based on the Matrox Imaging Library (MIL) to perform the routines that calculate the bottle’s coordinates, and determine whether the label is correct and affixed properly.
Because the winery needed to perform several inspections at once, CI Vision decided to use multiple cameras to capture images of the bottle’s entire surface. The 360 Full View has an aluminum enclosure around a conveyor that features a touch-screen operator interface on a side panel. As a bottle passes through the enclosure, it triggers four cameras simultaneously.
“Since the bottle is unoriented when it leaves the labeling machine, capturing the bottle’s entire surface was the best way to perform the label verification and bottle localization,” explains Rick Koval, lead software engineer at CI Vision. These two tasks involve several steps, and they all rely on MIL’s image processing functions. The system features eight fluorescent tubes mounted in a custom-designed fixture, four Basler GigE Scout cameras each with a 120° field-of-view, a custom PC with an Intel motherboard, and Matrox’s MIL software. In the near future, the system will also use the MIL GigE vision driver.
A bottle’s presence triggers the cameras, which each acquire a single image. Then MIL’s Measurement module determines the bottle’s location in space. When the bottle’s position is known, the system’s next task is to detect the locations of the front and back labels; the back label must be directly behind the front in order to pass the inspection. Warping functions and custom look-up tables “unwrap” the label image from the bottle, converting it to a two-dimensional equivalent.
With the flat images of the bottle, MIL’s Registration module detects the common areas in those images (the overlapping sections of the images), locates the edges of the labels with precision, and arranges them into a composite. The composite image is arranged so that the front label appears first, then the back label and, finally, another portion of the front label. This way, the user can see the entire surface of the wine bottle in a discreet image.
The labels’ positions are verified by their coordinates in the composite image. In other words, if the labels are placed at the correct height, their y -coordinates are equal. Likewise, if the distance between the labels in the composite are equal, then they are indeed directly opposite from each other. Furthermore, the composite image allows the user to perform other inspection tasks, such as checking for skewed alignment, and for the correct label. If, for any reason a bottle fails inspection, it is tracked and diverted through a reject mechanism.
All CI Vision systems use Matrox Imaging hardware, software, or both. “We like the Matrox Imaging Library because, frankly, it’s easy to use. Its C interface provides good low-level interaction with our own front-end software. And there’s a strong support team behind MIL that is very proactive and reactive to our needs,” says Koval.
The major challenge CI Vision faced was the development of the mathematical model for the “unwrapped” label in conjunction with applying the vision tools that take into account the random nature of consumer products. Glass containers often have slight aberrations and variations in the shape that affect the way the light refracts, or the way the labels are located on the bottle.
Although variability of the bottle’s initial position challenged one aspect of the system, it facilitated another. “Building a system that followed an existing process meant that we could use the existing conveyors—equipment that we were already familiar with,” says Koval.
The 360 Full View system is just one of a suite of products that Kendall Jackson uses on its production lines. CI Vision’s director of marketing Scott Stone added, “Flexibility and usability were key factors for us. We had to integrate very complex procedures into a system interface that allowed our customers to make changes very easily.” Training was also a factor. “We give customers step-by-step instructions on how to view the images on the screen, how to interpret the results, and how to change the piece being inspected,” he says.
Baby food jars, distillery products, and pharmaceuticals are all potential candidates for such sophisticated inspection systems, says Stone. He also foresees the day when automated systems inspect bottles from the empty glass bottle phase to the shipping-in-crates phase.
In the meantime, whether users are inspecting metal rolls, or delicate baby food jars, there will still be a need for high-speed, high-quality inspection. Machine vision hardware and software will continue to evolve to help machine builders compete.
|Renee Robbins is senior editor of Control Engineering. She can be reached at firstname.lastname@example.org .|