In the face of coming regulations, as well as forecasts calling for increases in food production and decreases in food waste, the task at hand for food and beverage processors appears to be a monumental one.
- Vision systems used in sorting
- Better detection and minimizing false ejections
- Vision systems for package inspection
- Quality assurance
- Reducing waste in sorting and inspection
To be in compliance with regulations such as the Food Safety Modernization Act (FSMA), food safety programs must have preventative controls in place to ensure the food produced meets US standards and is safe. At the same time, processors are also focused on increasing throughputs without sacrificing accuracy. Fortunately, a new wave of vision technology is helping processors automate their manufacturing and inspection processes.
New optical systems are designed to assist in reducing production errors, lowering manufacturing costs and increasing customer satisfaction while allowing higher throughputs without compromising accuracy. However, to select the correct vision system without wasting time and money in the long run, processors must first determine the results they want.
Farm to fork is a popular food topic in the mainstream media, but rarely, if ever, does the coverage take into consideration a very crucial step in that journey—the processing site. For processors taking raw produce from the field and transforming it into a quality product, sorting is absolutely critical. Consequently, those that previously relied on mechanical grading and manual labor are increasingly automating the process.
“Vision capabilities are much greater than the naked eye, allowing for fewer false rejects and better overall sorting efficiency on incoming defects,” says Eli Cannell, North America sales manager of food sorting for Sesotec. “Our customers’ bulk product is inspected with cameras from both sides, allowing nearly 360° vision. A software module identifies an optical contrast between good and bad product with our proprietary digital algorithms.”
Optical sorters can separate various grades of product according to visual characteristics such as color, size and shape. But which sorter is best depends on the processor’s needs.
“The first step for processors must always be defining their sorting requirements clearly,” says Charith Gunawardena, head of optical sorting for The Bühler Group.
To determine its needs, the processor must consider the commodity being processed, as well as the form in which that commodity is to be sorted, e.g., in shell, blanched, whole, milled or frozen. Then, the processor should determine what presorting steps need to be in place, such as de-dusted, sized or dried, and what input defects and typical contamination levels are expected. Next, the processor must determine the required throughput and the specifications regarding the final output’s accepted quality, capacity and yield.
Then, there is the question of where to place an optical system. “There are many positions in the line where you could put a sorter,” says Bjorn Thumas, director of market development at TOMRA Sorting Solutions. “It all depends on your primary resources, your incoming product and what kind of quality you want to deliver.”
For instance, if a product has seasonal differences with varying origins, a processor may decide to deliver several different levels of quality, depending on its end customers. Thumas says once the goals are determined, and if applicable, the existing production line is evaluated, several optical sorting platforms should be taken into consideration.
“When selecting an optical system, it is important to identify exactly what you need the machine to do,” says Steve Lizotte, business development director for Lizotte Machine Vision. “Do you only want to improve your quality control process, or do you want the optical quality and consistency available?
“A single machine will not have the same results for everyone. You want a reliable machine that not only grades or controls the quality of your product but also integrates into your production line perfectly so it does not slow down production,” continues Lizotte. “The bottom line is this: Look for a system that fits your production line, since some companies don’t offer the right belt speed, width or height. And look for a system that is fast enough to keep up with your production speed.”
Removing critical hazards and foreign material, whether they are physical, biological, chemical or allergenic, is a priority for all processors to ensure food safety. But, as Gunawardena says, “Improving the visual appearance of products also ensures value is added to the product being sorted.”
Previously, using automated sorters required balancing a tradeoff between higher product quality and a lower yield, due to potentially removing too much product that could be acceptable to the processor. However, according to Marco Azzaretti, advanced inspection systems product manager at Key Technology, “Recent advances in the technology are aimed at giving processors more options to manage product quality while simultaneously increasing yields.”
Unlike earlier equipment that primarily focused on rejecting defects, the latest innovations in optical systems are focused on creating a finer line of separation between good and bad. This allows processors greater control over the level of quality they want, which increases acceptable product while optimizing the product that does not meet the specifications for added value.
“These days, sorters aren’t added just for quality improvement, but for the benefit of yield improvement,” says Thumas. “For example, if you are doing X amount of pounds a year and throwing 3 percent away, and you minimize that to 2 percent, that 1 percent spread over a year could be a return on investment in six months for some companies.”
He explains that, although there used to be only a two-way stream, one for acceptable product and another for rejects that went to waste, there are now more options. By making a differentiation in the reject stream, a processor can improve yields and minimize waste. For instance, one stream made for misshaped or minor-defect products could be reused as Grade B with less stringent specifications, and another stream could be made for bruised product suitable for juices, flakes or cattle feed. “You’re not throwing away good product anymore,” says Thumas.
The latest software developments have focused on user-friendly machines that allow companies to control the specifications of rejections. “User-friendly systems significantly raise the level of flexibility for our customers,” Lizotte says. “In fact, this flexibility allows them to adjust grading parameters frequently and easily to meet each buyer’s individual demands.”
Thumas adds that while optical machines are becoming more intelligent, they also are becoming easier for operators to use. For instance, TOMRA machines include a Sort-to-Spec feature that allows the sorter to auto-sort for the color defects and size of French fries. The intelligent system analyzes the incoming product stream in real time through algorithms and measurements, makes piece-by-piece decisions and determines what is accepted and what is rejected.
The last decade has seen a rapid rate of innovation in optical sorting technologies. To help processors be more precise in the detection of small defects, one area of improvement has been the camera.
“The switch from analog cameras to digital was a big step in regard to image quality,” states Lizotte. “Analog cameras were prone to image distortion; with digital cameras, the image quality is always consistent and cleaner.” He says multi-chromatic technology also helps identify defects a monochromatic camera—one that detects only shades of gray from black to white—cannot detect.
“High-resolution cameras gather extremely detailed product data that is digitally processed for optimum contrast,” says Cannell. “High-speed camera technology has increased inspecting capacity to up to 7 meters per second. Resolutions as low as 0.09mm allow for unprecedented levels of defect detection.”
Additionally, the development of hyperspectral sensors is increasing the precision of detection. “For example, near-infrared [NIR] [technology] enables identifying the spectroscopic fingerprint of a specific material,” explains Cannell. “The product characteristics seen by the NIR technology are much more specific and unique for a material class compared to basic color sorting. Ultimately, this means better detection capabilities, less false detection and more accurate data.”
Azzaretti says Key’s proprietary BioPrint and Insort’s CIT systems combine broad-spectrum imaging hardware with intelligent software and algorithms to recognize objects’ biological characteristics. “Hyperspectral imaging allows detection of product conditions, defects and foreign material that are invisible on the surface of the object and undetectable by traditional cameras and laser sensors,” he says. “CIT, which Key has licensed on an exclusive basis to integrate on its potato sorting systems, detects ‘sugar ends,’ other invisible potato product defects and chemometric characteristics.”
Taking optical separation between good and bad product to another level, Bühler has developed the high-definition InGaAs technology, which makes it possible to clearly distinguish difficult-to-detect common foreign materials, such as plastics, wood and cardboard, in frozen vegetable mixes. “In rice sorting, we can enhance the distinction between subtly discolored grains and good grains by employing new lighting technologies such as textured lighting,” Gunawardena explains. “For other applications, we optimize the camera filters to distinguish between wanted and unwanted materials, and we use a combination of these technologies to remove a variety of different defects at the same time.”
Moreover, the high-resolution and high-power ejectors can open and close at very high speeds. They are controlled by sorting algorithms to ensure all targeted defects are removed, and the ejection is more precise.
Further down the processing line after the product has been packaged, vision inspection systems can play an essential part in a processor’s safety program, ensuring information on the packaging is readable and matches the contents, as well as checking seal closures.
“Vision systems can ensure the correct product codes are on a package, and that data can in turn be integrated into a larger traceability program—which keeps you in compliance with Hazard Analysis Critical Control Point [HACCP] protocols,” says Stephen Dryer, product manager for METTLER TOLEDO CI-Vision.
Vision system software can use numbers to identify a product or group of products. If a product is recalled, information on the package, such as the product’s name or its number/barcode, best before date and product date, and the producer’s name and address, helps in the discovery and control of the contaminated food. “If a recall is necessary, [processors] can identify the exact group that needs to be recalled based on the assigned number,” says Lizotte.
“Vision systems can inspect tamper band and seal integrity, and verify the presence of expiration dates and other tracking information,” says Dryer. “These capabilities not only ensure food is safe, but in the unfortunate case of a recall, they can help make conducting that recall much easier and more effective.”
In recent years, 40 percent of food recalls have been due to the presence of an undeclared allergen, he adds. For a person highly allergic to a substance, even a small amount ingested can be fatal, making allergen information critical for food safety. Vision systems can check whether allergen information has been applied correctly.
“By verifying the correct label is on a product and that all codes and text are present and readable, a vision system prevents mislabeling and other labeling errors that can result in a product recall,” says Dryer.
Seams and seals that are not properly closed can lead to spoiled product and unhappy consumers. They can also damage to a brand’s image. “Our vision systems are used for quality assurance—everything from ensuring a package is free of holes or tears to making sure the right label is on the product,” states Dryer.
Dansensor recently introduced a new vision system called the VisioPointer packaging inspection system, which uses software to immediately locate product that has been trapped in the sealing area. “The technology development has improved the ability to achieve close to a 100 percent fool-proof sealing inspection,” says Steen Andreasen, product manager for Dansensor. It can also detect smaller errors with fewer false rejections while working at an increased speed.
Similar to the improvements in sorting vision systems, package inspection equipment is making advances in camera and lighting technology, as well as user friendliness. “Better software has not only improved the accuracy and performance of inspections, it has made advanced machine vision more accessible,” says Dryer.
Ease of use was a central consideration in the development of the METTLER TOLEDO CIVCore vision inspection software platform. “We listened to our customers and incorporated what we learned into the new interface, with the most important information and settings only a tap or two away and deep functionality for power users that is just as easily accessible,” says Dryer.
The software allows administrators to set and adjust tolerances for each inspection. The ability for a user to do this or not can be controlled through a software log in; all modifications are recorded to keep track of changes.
Even though optical systems used in sorting and inspection processes serve different functions, the processes have one goal in common—reducing food waste.
“Our customers are very conscientious about the scarcity of food,” says Thumas. “In the past, sorting was just about taking out the defects. Today, it’s a two-pronged approach that aims not to only achieve the quality the end-consumer expects, but also to increase the amount of end product and optimize the reject stream product, so that it’s not just waste. Waste costs everybody money.”
This is also the objective of the Bühler SORTEX E BioVision machine, which was developed for almond producers that have traditionally had a problem removing pale shells that are similar in color to paler varieties of almonds. The solution targets the spectral and spatial difference between the nut meats and shells, which allows subtle differences to be distinguished. According to Gunawardena, “Without this technology, a much higher volume of the nut meat would be lost due to false rejections.”
On the packaging inspection side, vision systems that check labels and seal integrity can cut down on the amount of returned or scrapped products by identifying a problem before products leave the processing site.
“They can reduce the quality cost related to scrap and, equally as important, reduce the cost related to the shipments that are rejected by customers due to wrong or improper labels and printing,” says Andreasen.
Improperly sealed food can drastically reduce shelf life by allowing protective gas to escape from the package, which could potentially lead to product spoiling before it even reaches the retail shelf. If seal integrity is ensured, more products can avoid becoming garbage and instead contribute to feeding the world’s growing population.
For more information: