Machine vision ensures perfect dairy packaging
Vision technology has a crucial role to play in offering manufacturers in the food industry a competitive advantage. One important area is packaging, where both the quality of the packaging itself and the markings on it need to be checked. TriVision, based in Odense, Denmark, has developed two vision-based systems featuring vision components supplied by STEMMER IMAGING, which provide 100% inspection of packaging at the manufacturing stage and of the final product. These are in use with leading companies in the dairy industry such as RPC Superfos and Arla Foods.
Plastic tubs and their lids for dairy products such as butter, cheese and other spreads are generally produced using the in-mould labelling (IML) process. Here paper or plastic labels are used as an integral part of the blow moulding, injection moulding, or thermoforming processes. This allows the final container to be produced complete with artwork. Any containers with structural defects arising from the manufacturing process, such as burrs, variations in surface finish and variations in size must be rejected. With the IML production process, the integrated artwork and barcodes also need to be checked for correctness and placement.
When acceptable containers reach the dairy and are filled with product and sealed for dispatch, further checks are also essential. This not only helps to prevent poor end-user perception but also could avoid product recall that might result from incorrect labelling. This could be both costly in a manufacturing sense and very damaging to a brand’s reputation.
Developing a vision solution
The ability of industrial vision systems to make a wide variety of repeatable measurements at high speed makes the technology an essential tool in manufacturing.
TriVision A/S was established 1999 to concentrate on vision technology solutions for the food packaging and food production industries and to help customers gain higher efficiency and quality in production. The Packaging Inspector was introduced for inspection of the containers at the manufacturing stage, while the Product Inspector is used on the final packaged product. Launched in 2007, the Packaging Inspector was developed in collaboration with RPC Superfos, a leading designer, developer and manufacturer of innovative plastic packaging solutions.
The requirement for the system was 100% inspection with all bad products rejected. Working closely with RPC Superfos allowed TriVision’s vision experts to fully understand the range of defects that could occur during production of the tubs and to develop the appropriate imaging solutions.
TriVision also has a strong working relationship with STEMMER IMAGING, which stretches back to 2005. STEMMER IMAGING is Europe’s largest independent supplier of vision technology and services to industrial and scientific OEMs, system Integrators and reseller markets. STEMMER IMAGING supplies all of the cameras used in the Packaging Inspector and Product Inspector systems as well as camera cables and some of the optics used on the system. The company also supplies line scan cameras and frame grabbers to TriVision for other food inspection applications and has worked with them on a hyperspectral imaging project.
For the Packaging Inspector it was important to use cameras with good spectral sensitivity and the ability to detect very small moulding errors. STEMMER IMAGING was able to provide advice on different cameras, especially in terms of the image quality produced and the on-board processing capabilities.
Cameras from Allied Vision were selected and in the early versions of the inspection products the IEEE 1394b (Firewire) interface was used, Later versions were able to take advantage of the higher data transfer rates and longer cable lengths offered by the GigE interface standard, with Allied Vision’s Manta and Mako cameras being used. In addition, the use of GigE Vision cameras allowed cost savings since frame grabbers were no longer required.
Ole K. Neckelmann, Commercial Director at TriVision, commented: “Our relationship with STEMMER IMAGING has been invaluable in keeping us up to date with all the latest models of cameras that come to market and whether they would offer us any additional benefits for our inspection products.”
TriVision has taken responsibility for developing all of the software for the products internally, making use of Allied Vision’s API. A number of methodologies are used, including pattern recognition to check for the presence of labels and artwork.
The inspection process
In order to make 100% inspection with multiple measurements, a number of cameras in different positions are used. The Packaging Inspector is used to inspect manufactured containers for the following:
- Overmould errors
- Dosing errors
- Holes and creases in labels
- Artwork errors
- Positioning of labels and artwork
- Dimensional measurement of the containers
- Analysis of the barcode
The primary inspection is for overmould errors. These occur when the injection moulding process fails due to small amounts of liquid plastic running on the wrong side of the label. When this happens, the plastic container has a leak.
Dosing errors occur when there is too little or too much material in the container and always show on the edge of the product. For underdosing, material is missing on the edge of the product. For overdosing, there is extra material on the edge.
For rectangular containers, two cameras are positioned below a conveyor belt to inspect the box as it passes above them. Simultaneously, the printed artwork is inspected from above to check the correct position of the label, barcode, etc. The lids used to seal the containers are inspected on separate lines. The inside surface of cylindrical containers is inspected from above while a camera mounted laterally checks the presence of the label on the outside surface.
The Allied Vision Manta and Mako cameras that are used benefit from a wide range of on-board image optimisation features and offer flexible camera control. Cameras with resolution ranging from 0.4 Mpixels to 5 Mpixels are used depending on the particular inspection task and the tolerances required. For example, extremely small moulding errors need to be detected at relatively high speed. Measurement tolerances are specified by the user both for the particular type of measurement being made and the individual product, but dimensions are typically assessed for 0.1 mm variation.
The software offers great flexibility. In most installations, there are frequent manufactured product changes and the system is able to quickly learn the required parameters for a new product. In addition, the data from each measurement is used to analyse the status of the production procedure so that early intervention can be made if defect trends from a particular product line are identified.
Packaging Inspector in action at RPC Superfos
The TriVision Packaging Inspector is a modular system and can be adapted to different products and tasks. RPC Superfos A/S uses Packaging Inspector in IML manufacturing to detect overmoulding errors in assemblies and corners of the containers, provide control of barcodes and detect artwork errors. These include wrong or missing labels, variations in text and/or colours, incorrect positioning of labels and labels with holes, creases etc.
There is full user specification of tolerances for rejection criteria such as the size and contrast of overmould errors, the amount of label misplacement, the size of defects due to under and overdosing, the length and width of the products and the evaluation of barcodes. The simple teach in of parameters for new products can be completed in less than 5 seconds.
The system also counts the number of ejected items to prevent the ‘reject’ bin receiving more than the maximum number specified by the user. In addition, any partially filled reject bin has to be replaced and confirmed by the user at product changeover, to avoid mixing of reject box types.
The Packaging Inspector is equipped with a graphical real-time statistical function, where every error is logged against the specific cavity tool. This allows the operator to check the status of each tool and quickly identify the type of error detected. The system also comes with fast and direct online support by remote desktop for updates and each system can be supported by a VPN. This means that TriVision can connect to directly to the production line if there is an issue that cannot be resolved locally. IML production methods are constantly evolving and the Packaging Inspector provides the flexibility for future developments.
Product Inspector in action at Arla Foods
The Lurpak & Arla Kærgården brands from Arla Food are known world-wide as leading products with high quality and taste as well as pack appearance and design. Arla wants to deliver perfect products to their customers, so they use TriVision’s Product Inspector to ensure the tubs have the correct barcode; the lids have the correct number; that there are no flaws in the artwork; that the printing is correct and in the right place and that the protective paper is correct and accurately positioned. All tubs that are not 100% perfect are removed from the production line.
The Product Inspector system features a cloud-based Production Intelligence Module, which makes sure that the production equipment is always in optimal condition. This reduces the number of stops in production and increases the overall efficiency. If a major or minor malfunction occurs in the production equipment, the operator can quickly identify the problem from the module and even get assistance as to how to rectify the fault.
The system allows the quality of different products and different production lines to be compared and makes sure that the production line has more good products per hour than a similar line without the inspection system. The number of customer complaints has also decreased significantly.
TriVision continues to develop its product range and with the help of STEMMER IMAGING, monitors camera availability to ensure optimum performance and value for the customer.
Ole K. Neckelmann concludes: “Our systems have been highly praised by our customers, and several systems have been re-ordered. They are characterised by their ability to handle natural variations, the user friendly interface, the simple teach-in of new products and the analytical capabilities based on data captured during the production process”.