Machine Vision in the Container Industry -- Part 2: Metal Containers, Plastic Containers and Closures

In the first part we reviewed the applications and players in the glass container market, both for glassware manufacture and filler line applications. In this article we cover the balance of the applications for machine vision in the container industry: metal containers, plastic containers and closures.

Metal Containers

In the metal container industry, the applications are largely found at the manufacturer although there is no reason why in general the same systems could not be used on filling lines. At least one company is offering such a capability.

One of the earliest applications of machine vision addressed in the early 80's was inspecting the sealant region of a can lid for a two-piece can. While driven by the need to verify presence and integrity of the sealant, the expectations for machine vision system ultimately became to inspect the product side of the lid comprehensively for all other potential reject conditions. Vision systems typically addressed these requirements by looking for gray shade changes radially and/or circumferentially around the lid. Today machine vision systems inspect for conditions such as mechanical deformations, missing coating, compound voids and skips, cut/bent curls, die marks, ink or grease, foreign objects, dents, holes and scratches. All this at speeds up to 1200 per minute.

Looking at the consumer side of a can lid became important with the adoption of pull-tabs on beverage cans. In this case, the functional regions associated with the pull-tab are critical, so dimensional checks were required as well as pattern recognition-based feature analysis. Today machine vision systems can typically check the following dimensions associated with a can end: end diameter, curl diameter, curl height, curl opening, tab orientation, rivet diameter, etc.

Ultimately, especially in the beverage market, as designs emerged permitting ever-thinner walls, a variety of inspection requirements emerged to inspect the cans themselves. These include both dimensions and defects that can be seen as anomalies in the texture. In the case of the can itself, machine vision systems can typically make the following dimensional checks: finished can height, trim height, flange width, flange angle, plug diameter and eccentricity, bead height, seam width and height. These systems inspecting 100% of the product produced are able to provide real-time statistical data related to present production processes to assure corrective action is taken before conditions result in rejects.

Anomalies machine vision systems inspect for include those along the sidewall of the can as well as at the bottom. Being able to inspect along the entire wall given the height of a typical can represents a challenge that has been addressed by a number of suppliers with creative lighting/optics/camera arrangements. These machine vision systems are designed to detect conditions, such as ink or grease stains, neck puckers, dents, holes, scratches, missing coating, coating bubbles/blisters, excess lubricant, inside litho transfers, split and knock down flange, slivers, wrinkles, etc.

Today there are some products that can inspect the outside can deco or can labels for color integrity and other printing concerns as well as correctness - also called 'tramp can inspection.'. At the same time they can inspect for dents, wrinkles, inclusions, metal flaws, lacquer flaws, etc.

While some of these capabilities have been around since the early 80's, today the machine visions systems that are on the market addressing dimensional and attribute/flaw inspection applications in the metal container industry are substantially improved. While in general the creative lighting/optics/camera designs are what make them capable of addressing these applications, the components themselves have become substantially more robust for the applications over the years.

Today telecentric optics makes it possible to provide consistent and repeatable images not influenced by magnification changes stemming from depth-of-field issues or vibration issues. Many of these systems employ application-specific LED lighting designs that incorporate directional strobing designs to optimize the sensitivity to flaws in specific regions. Geometric-based flaws create specularly reflected or diffusely scattered signatures depending on the specific lighting design. Correspondingly, those characterized by being specularly reflective can be distinguished and classified distinctly from those characterized by diffusely scattering light.

Back in the early 80's the biggest challenge when dealing with high-speed lines such as found in the metal container market was the camera. Besides typically being a vidicon-based design with all their inherent instabilities, the cameras were all operating at 30 Hz. How to synchronize a field (half a frame and 60 Hz) with a strobe and with image capture was a real challenge. Most of the early systems essentially sacrificed vertical resolution by only using a field of data. Horizontal resolution was also sacrificed, as the field-of-view generally had to be increased to compensate for the positional uncertainty in the field captured as a consequence of synchronization uncertainty. Today, higher resolution cameras are also being applied for these applications, leading to increased sensitivity to concerns.

All these problems have virtually disappeared with the cameras that are now available. Using asynchronous scanning mode permits full capture of each and every scene event and relatively easy synchronization of strobes and image capture cycles. Progressive scanning permits capture the entire vertical resolution. With precise synchronization it is no longer necessary to open up the field-of-view to assure the image is always in the scene so the full resolution of the camera can now be applied horizontally.

With all these advances in the underlying system components as well as the advances in compute technology, it is now possible to provide systems totally responsive to the requirements of the metal container market. Today speeds up to 2400 per minute are also possible.

Plastic Bottles

As with glass bottles there are two distinct markets: plastic container manufacture and container filler. It turns out that for the container market there are two segments: preform manufacture and ultimate blow molder. Not all plastic container manufacturers are vertically integrated. Inspection of the preform is tied to the specific cavity producing the preform so that faulty cavities can be corrected quickly. The preforms are inspected for conditions, such as short shots, nicked finishes or oval finishes, body diameter, total length, gate length and straightness. By inspecting the preforms as they are manufactured and reducing the number of preforms with sealing surface defects, blow molder efficiency is improved by virtually eliminating the incidence of damage to heat lamps or jams due to defective preforms.

In the case of the plastic bottle itself, inspection can be performed inside the blow molder or on an external conveyor. Just as in the glass container applications (in fact most of the companies offering glass container inspection systems also offer systems to inspect plastic containers), modules exist that can inspect the various areas of the container: base, neck, seal surface and finish.

In the base, inspections are generally performed on a regional basis permitting gate, fold and petaloid base inspections. In the neck area, machine vision systems inspect for contamination, folds, process rings, choked neck, etc. In the sealing area machine vision systems inspect for defects such as nicks, dents, split finishes and surface irregularities. In the case of in-mold labeling, machine vision systems can inspect the container for label integrity, skew, position, folded corners, etc.

Systems inspecting preforms and blow-molded bottles can operate at up to 800 per minute.

Closures

One early challenge for machine vision was the requirement to inspect an aluminum closure after it had been applied to the beverage bottle. The closure itself was a simple cylindrical sleeve that was dropped onto the bottle. Rollers, essentially rolling over the threads of the bottle itself, formed the threads. If the pressure was too high, the rollers could score the metal. If the pressure was too low, the thread depth would be too shallow. Either way, in the case of a gasified beverage, when a person went to open the bottle, the cap could release prematurely. This had resulted in a number of lawsuits as consumers complained that the premature release of the cap resulted in bodily injury.

Taking on this challenge in the late 70's was one of the pioneering machine vision companies - ORS Automation. Devising a clever prismatic optical arrangement and utilizing one of the earliest CCD cameras, and very early stage microprocessor technology, a solution was beta site tested operating at 1200 bottles per minute. An early lesson in marketing learned was to pay attention to concurrent industry developments. In the two years it took to develop the system and prove its functionality, the plastic industry was busy coming up with plastic closures. Not that the aluminum closures disappeared overnight, but the availability of an alternative to the aluminum closure for capping gasified beverages, (along with the cost of a system) resulted in a loss of interest in the solution.

Today, the machine vision inspection of caps, crowns, closures, etc. at the manufacturer is rather commonplace. The objective of these systems is to inspect for anything that could compromise function and appearance. Functional defects include:

• The closure's proper attachment to the bottle
• The sanitary/pressure seal (if applicable)
• The tamper-evident features (if applicable)
• Smooth operation of the filling/capping equipment - avoidance of jams, torque errors, etc.

Appearance defects include:

• Print quality on printed caps
• Proper cap/liner colors
• Flash or other irregularities in the shell or consumer side of the closure.

Conditions these machine vision systems inspect for on the product side include liner missing, liner voids, plastic voids, out-of-round, cut outs, knock downs, deco verification, contamination, wing/flute defects, etc. On the consumer side machine vision systems can perform color-based inspection of the litho, as well as check for many of the cosmetic concerns cited as on the product side but which can also be found on the consumer side. These systems can make these inspections at rates up to 3000 per minute.

The tables reflect suppliers of application-specific machine vision systems that address the applications described herein. They also reflect those that are known to offer products for these specific applications in the North American market. There are other companies that offer similar application-specific machine vision products, but so far they are not known to have participated significantly in the North American market. It is also observed that there are any number of merchant system integrators who have also delivered systems based on general-purpose machine vision systems, frame grabbers, embedded vision processors or smart cameras for these applications, typically on a project-specific basis. Since they are not perceived to have commercialized their solutions, they have not been included.