Don Braggins, UK Industrial Vision Association, highlights developments in inspection technology and visions systems for use in pharmaceuticals production
Industrial vision is a well-established technique in the pharmaceutical industry and its applications continue to be primarily in quality control. These can range from checking fill levels in vials and bottles and the presence and position of tamper-evident seals to security and traceability applications such as checking the presence or absence of labels, character recognition (for example, date code recognition or product type) and print verification.
Vision systems bring real benefit to the production line by virtue of their ability to make repetitive measurements accurately, at speed and objectively. Vision systems also remove human frailties from the inspection process since they are equally vigilant at the end of a 10hr inspection shift as at the beginning.
The cost of a single product recall is probably going to be orders of magnitude greater than the cost of several vision systems, and no matter how many human inspectors are used, they can never be as untiringly watchful.
Recent technological developments, such as vision sensors and the GigE Vision image data transmission standard, together with improvements in smart cameras and code readers, have not only made it much easier to interface vision systems into pharmaceutical production lines, but have also improved speed and performance.
Vision sensors are a new class of inspection products that combine some vision capabilities with the ease of use, fast deployment and high-speed binary output of traditional photoelectric sensors. Vision sensors offer more than simply sensing the presence of an object such as a bottle lid. Some detect and inspect parts on production lines or check for the presence of specific features on products and packages. Others perform tasks such as synchronisation, profile detection, object recognition or height measurement.
plc capabilities
Some vision sensors do not limit the number of areas on a part or package that can be inspected, and include PLC-like capabilities to provide one single pass or fail result. This typically requires that the vision sensor incorporate a processor and memory. Some more advanced vision sensors accept encoder input directly and include an automatic shift register to track and reject items on variable speed packaging lines.
Finally, many vision sensors can store multiple set-up files for various products that are selectable by discrete inputs for very fast line changeovers. As an example, AstraZeneca in China uses a vision sensor system with an integrated LED illumination and lens on a label production line (see Figure 1) to check whether the edges of adjacent labels are properly aligned, as well as counting the labels.
The system sounds an alarm to alert the operator if an error is detected to ensure that there is a zero defect level for label presence and positioning.
smart cameras
A smart camera is a self-contained, stand-alone vision system with image processing capabilities built into the housing of the camera, removing the need for a local PC. Equipped with appropriate communication interfaces, e.g. Ethernet for programming and distribution of results, they can have industry-standard I/O lines for connection to a PLC, actuators, relays or pneumatic valves. Smart cameras feature powerful image processing tools for a wide range of vision tasks.
An alternative approach is the compact vision system, where all the processing power is contained in a rugged, DIN mountable enclosure, instead of the camera head. This allows much smaller cameras to be used and multiple cameras can be connected to the single processing unit.
Developments in smart cameras have resulted in reduced size even with rugged housings (Figure 2), meaning that there is more flexibility of positioning within the production environment. Although smaller smart cameras are available, image resolution continues to increase and as the processors used within them become more powerful, both the range of applications available and the speed at which images can be processed (and hence the rate at which product inspection can take place) have increased.
Smart cameras are in use at Fresenius Kabi for high-speed inspection of bottles of serum for impurities as small as 1mm2 at a rate of 10,000 vials per hour. The vials of serum are contained in a blisterpack, with five vials in each pack. The vials pass along a conveyor where they are inspected by two smart cameras (Figure 3).
The on-board vision tool is configured for part finding and localisation to locate the units of serum so each individual vial can be inspected for impurities. The detection of any impurities results in the entire pack being rejected. All inspected vials that have passed are then sent for packaging. The vision solution is fully integrated into the company's quality management system.
The smart camera is a low cost, low risk way of getting people to use vision for the first time. Prior to the introduction of GigE Vision technology, the smart camera approach overcame fundamental difficulties in being able to transmit high quality images over long distances in a production environment by processing the image inside the camera and simply transmitting the result.
The recent introduction of the GigE Vision and Gen<i>Cam standards, developed by groups of companies from within the vision industry itself, has made a significant improvement to the digital transmission of image data.
GigE Vision is a camera interface standard developed using the Gigabit Ethernet communi-cation protocol, which allows image transfer at good speed using low cost standard cables and components over very long lengths, so that the processor can be housed away from the production line environment. Compatible with a wide selection of industrial-standard connectors, cables and components, this technology can readily be integrated into existing Ethernet-enabled production environments.
With GigE Vision, hardware and software from different vendors can inter-operate seamlessly over GigE connections. Gen<I>Cam is a central software interface for controlling cameras, no matter what interface technology (GigE Vision, Camera Link, Firewire, etc.) they are using or what features they are implementing.
code and labels reading
Correct packaging labelling is critical for consumer safety in the pharmaceutical industry. Coding and labelling usually takes the form of alphanumeric codes (e.g. lot details and best-before information), barcodes and the more recently introduced 2D codes. Products can be tagged either by a stick-on label or by information printed directly onto them or onto the packaging. Machine vision methods for reading codes and labels include pattern recognition and matching, optical character recognition (OCR) and 1D code reading. The challenges faced by the vision system include:
- random orientation of the product giving random orientation of the code or label
- position of the code on the product (irrespective of orientation)
- codes or characters that deviate from the norm (e.g. stretched or skewed characters)
- contrast between the code or characters and the background
- code/background colour combinations that may not be acceptable
- factors such as reflections from surrounding surfaces
- physical space in the production line for the vision system
- interface between the vision system and a reject mechanism
- and, last but not least, the speed at which the measurement can be made.
An alternative is the Data Matrix code, a 2D code with a square or rectangular array of dots, or cells, that packs much more data into a far smaller space. The 2D code can be read with very little contrast, typically down to 20% or less, compared with 80-90% contrast required for 1D barcodes. As the data in the code is encrypted with an Error Correction Code, up to 20% of the cells within a code can be destroyed or defaced and all the data can still be read without error, making the code able to withstand handling damage.
Recent developments in 1D code reading technology, however, have significantly improved 1D code reading capabilities. Although 1D barcode scanners could generally identify damaged codes as "unreadable", a new generation of compact 1D bar code laser scanners (see Figure 4 opposite), offer improved reading algorithms with a SMART decoding system for rapid and reliable reading of even badly damaged or partially covered bar codes. These also have Ethernet connectivity.
About UKIVA and its supplier members
The UK Industrial Vision Association (UKIVA) is a not-for-profit organisation, whose prime objective is to promote the use of vision by manufacturing industry. Many of the Association's members are involved in the supply of vision systems and components to the pharmaceutical industry. The UKIVA web site (www.ukiva.org) has been designed to be a machine vision information hub. It is the first port of call for many who are seeking to introduce a vision system.
The pharmaceutical industry was an "early adopter" of vision and it was a conversation at an exhibition for the pharmaceutical manufacturing industry in 1991 that led to the formation of the UKIVA in the following year.
When UKIVA was formed, an eventual target membership of perhaps a dozen companies was envisaged â€âEuro Å" there are now more than 30 members.
During its first few years of existence, the UKIVA was administered by a larger trade association, the Process and Packaging Machinery Association (PPMA). After 14 years of independent operation, discussions are now under way for the Association to become a "special interest" part of PPMA while retaining its identity and website, alongside the British Automation and Robotics Association, which made a similar arrangement earlier in 2009, and this should lead to an even stronger grouping of suppliers of automation systems and components.
