The sealing surface and top inspection of glass containers has a long and important history. The earliest techniques of inspection were very basic and mechanical in nature; inspecting for dips, saddles, tilt or chips that can affect the seal of the container. Over time, the sealing surface and top of the container evolved into one of the most important areas of inspection for quality and consumer safety. As a result, camera-based techniques were developed and initially introduced with a top view for inspecting the sealing surface for defects. Some of these inspections did not require rotation (non-contact) of the container, while others did require rotation and contact.
Continued advancements in forming process technologies and mould designs, combined with increased speeds, provided the glassmaker with cost-effective and flexible methods for producing containers. However, these advances also increased the need for post process quality assurance, in order to reject defective containers and provide process-related feedback to the hot end, in order to correct problems. Defects such as wire edge, over press, chipped and broken finishes, blisters in the finish area, line overs, knockouts, unfilled finish, crizzled finish, offset finishes, out-of-round, seeds and split finishes all require optimal lighting geometry, lensing and optics and camera resolution within the inspection station.
Optimal lighting geometry
Each of the above-mentioned defects has a different optimal lighting geometry and optical configuration specific to the defect, in order to maximise detection and minimise false rejection. This conflict of illumination and optical arrangement has been a limitation of traditional vision-based inspection and in particular, sealing surface.
Illumination source and illumination geometry had to be developed in such a way that as many defect types as possible are detectable on the sealing surface, with some being very detectable, while others less so depending on location on the surface and type of defect. Some defects like wire edge and over press have illumination requirements and camera angles so different from traditional sealing surface lighting and optics that a separate inspection module was necessary.
Until now, this had been a limitation of traditional vision-based providers. The user was faced with finding a compromise. For example, optimal lighting and lighting angles, camera parameters, lensing and optics for locating the sealing surface are a suboptimal configuration for line over detection. Conversely, optimal lighting and angles for line overs may be suboptimal for knockout inspection.
Clearly, all sealing surface defects are important but with monochromatic lighting paired with cameras and existing lighting geometries, operators often had to compromise the detection of one defect, while looking for another. The operator had to optimise lighting and settings for the defect of interest, while trying to minimise the negative consequences on the other inspections being performed. These compromises resulted in more difficult and complicated inspection setup for users. The result of these compromises is less detection capabilities on the sealing surface and/or more false rejects.
Imagine holding a beer bottle with multiple sealing surface defects using the light in the office as the illuminator and an eye as the camera. When looking at the bottle and moving it relative to the light at different angles, the defects will go from more visible to less so, depending on the defect geometry and lighting angle.
Each defect has optimal lighting angles for its detection. Defects with geometries such as wire edge, over press, dips, chipped finishes and knockouts change the light reflected back to the camera because of their geometry. As a result, these defects become very recognisable. Unfortunately, in a production environment it is not practical to perform this type of container manipulation quickly at full production rates, so compromises had to be made.
A different approach
Applied Vision has never settled for compromises like these and saw an opportunity to create value for the glassmaker by developing a technique for sealing surface inspection that was different from any existing approaches. The target to improve detection capabilities, decrease false rejects while combining as many defect inspections as possible into the sealing surface module was achieved using Applied Vision’s patented Multiview technology. Mutliview utilises the latest colour camera technology, combined with a multispectral illumination geometry. Applied Vision also successfully integrated the wire edge and over press inspection into the same inspection module without the compromises mentioned previously. This revolutionary approach utilises a fixed geometry of multispectral light emitting diodes (LEDs), controlled by the company’s software. The camera utilised for acquiring the image is a high resolution colour unit, with specialised optics at a fixed distance and angle to the sealing surface. With Multiview, the contours and shapes of the object being imaged result in different colours of light being reflected back to the camera. From one single colour image, a nearly infinite number of filtered colour images can be utilised for each specific defect inspection.
Acquiring the image and applying the appropriate colour filter to that image are only the first steps of performing the inspection. Software tools used to locate the sealing surface in the image are the foundation of the inspection.
Applied Vision also developed the Adaptive Locator software tool, which essentially learns, locates and registers the sealing surface in the image automatically, taking the burden of location setup away from the operator. Once the image is registered, Applied Vision has a large library of proprietary algorithms that can be utilised for inspecting defects in these processed images. The difference with Multiview is each tool, or targeted inspection is assigned to a colour filter that provides the best contrast and signal for detection. This results in less false rejects.
A tremendous amount of processing is accomplished in a short amount of time due to these fast and effective inspection algorithms. The use of colour to inspect the sealing surface has greatly improved the ability to reject and provide feedback to the hot end. Each image provides valuable information for effective defect classification.
Applied Vision offers this inspection module in both its Volcano SSB sealing surface inspection module, as well as Cyclops, a stand-alone sealing surface and wire edge inspection system.