Image processing cameras check beer crates

At the Furstenberg brewery, more than 3000 beer crates per hour undergo thorough quality checking, while they are still wet after cleaning, using a machine vision system and four cameras.

In the brewery business, the quality of the product plays a very important role, which is also defined by strict legal regulations.

Brand image is very important and every small contribution to it improves market success in a difficult competitive environment.

Product presentation therefore needs to be immaculate, to ensure this at the Furstenberg brewery, more than 3000 beer crates per hour undergo thorough quality checking, while they are still wet after cleaning, using a machine vision system and four cameras.

This unique turnkey solution was implemented in a partnership between Cognex, the world's leading industrial vision company and Siemens Industrial Solutions and Services, a systems integrator.

The inspection station has been working for over a year without a failure, and guarantees that only perfect beer crates reach the shops.

At the Furstenberg Brewery, beer crates are reused, and may be handled very roughly in daily use.

Damaged or dirty crates reaching the customer are unacceptable as they could leave a negative impression, which is often registered unconsciously, and may have an effect on the next purchase.

Correct appearance of the goods is important, for instance, a price label left on the beer crate from an earlier special offer can annoy both the seller and the customer.

In the past, damaged or dirty beer crates were constantly slipping through the labour-intensive manual checking, and creating complaints.

"We looked for an economical way of solving these problems, or at least minimising them.

We also wanted to achieve a constantly updated statistical overview of the whole stock of beer crates", said Helmut Bartels, production and engineering division manager at the Furstenberg Brewery.

In contrast to the existing experience with image processing systems in the bottle cleaning and filling plant, for this task there was not yet a complete system solution on the market.

It was quickly established that the requirements of the specification could only be met with a very powerful image processing system.

The search for a system solution turned in to an involving project.

The selection check had to achieve a whole series of features for the rejection level, with high reliability, for different types of beer crate.

It was important to integrate the checking station into the existing conveying equipment without extensive rebuilding.

This required a particularly space-saving system solution, which had to ensure reliable functioning in spite of sprayed water, water vapour, changing conditions of room lighting and wet beer crates.

It was also important to achieve the aim of about one beer crate per second on the conveyor.

Checking with several high-resolution cameras simultaneously and looking for a series of different features required a lot of computer power in the image processing system.

Finally, the existing personnel would have to cope independently with adjusting the selection.

Of the four crate types in common use, in the first stage the two newer types would be checked for reusability depending on visual appearance and mechanical defects, and automatically rejected if necessary.

The following detection criteria were required: dirtiness of minimum 1cm2 dirt, glue, remains of labels; damage to imprinted writing, coat of arms or crate size data; and mechanical defects such as breaks, chips.

"On the search for a vision supplier, there was no ready-made solution for this requirement.

Cognex, with the Checkpoint II system, was the only company that could demonstrate potential for fast implementation on the basis of a laboratory setup, meeting all inspection criteria.

Because of the support of Cognex as a vision partner, and its great experience and know-how, we were able to get into the actual project phase very quickly", said Martin Sluzewski of Siemens Industrial Solutions.

The success and reliable checking of this application is based not least on the capabilites of Checkpoint II.

Distinguished by its powerful hardware and software and with four simultaneous cameras Checkpoint II copes with even time-critical, complex image processing requirements.

The performance of this vision system is based on a high-performance board with four camera inputs, an integrated 400 MHz vision processor, its own working memory, optimum support of the Intel MMX instruction set and the unique PatMax technology - a software tool for detecting the position of objects by imitating their basic geometric structure.

For space reasons, it is not possible to view all four faces of the beer crate square on.

Some of the cameras view the crate at an angle, giving a distorted image, but because of the powerful Cognex image processing tools, it is still possible to reliably inspect each face of the crate.

The complex patterns in the Furstenberg logo, the coat of arms and the beer crate content label are analysed by PatMax.

The threshold, with a sensitivity range of 0 to 1000, defines the point from which the pattern is recognised as bad.

The "blob tool" ensures that foreign bodies on the crate are found.

The "minimum blob size" setting (in pixels) and grey value threshold parameter define the point from which a foreign body affects the analysis.

The "minimum edge gap" parameter defines how many pixels must separate a pair of edges for it to be defined as a fault.

The "light meter" vision tool, with parameters, detects foreign bodies and defects by the average brightness of an area, and the flags on the coat of arms are compared to a stored reference image.

Checking all these many features over four cameras for each beer crate reliably takes only about 800ms.

Another important aspect for the brewery was that employees can make adjustments to this vision application themselves, very easily and at any time.

This is enabled because of the very simple user interface produced using Microsoft Visual Basic.

The crate types with their data and reference images are stored in a library, and can be activated by a mouse click.

All programming is taken away from the employees, since they can make all adjustments simply by changing parameters.

In addition, by studying the statistical data from the system it is possible to order new beer crates to match demand.

The simple user interface proved itself in the testing phase in realistic process conditions.

The functional reliability of this system solution was also shown in the final installation phase, for which only 3 days were required.

This was immediately followed by normal three-shift operation.

The positive experience of one year of fault-free operation is already reflected in the current extensive expansion of production, because another checking station of the same kind is already planned.

Georg Catterfeld, who is responsible for this project in the brewery, commented: "We were very impressed by the quick project implementation, productive capacity and reliability of this system solution".

"This very complex task can be seen as the pilot project for similar applications.

The partnership and vision know-how of Cognex were decisive for the success of this turnkey system solution", was how Martin Sluzewski of Siemens summarised the experience.

Back to top