Bottle Inspection with Vision Sensors
The global beverage industry is growing constantly and the people are spending more and more money on it. In 2013 people in the whole world spent 3 trillion EURO for drinks and packaged food.
The growing demand requires a more efficient production. At the same time also the demands on quality are increasing. Therefore the manufacturers are forced to invest into faster and better machines and plants respectively. In the beverage industry this already starts with the production of the preform of a bottle and repeats itself, when the empty bottle is recycled and has become a new bottle.
“The automated visual quality inspection in the food and beverage industry is unstoppable” says Michael Beising from EVT, “there are no limits in what can be visually checked.” The ways of inspecting the content and the packaging of bottles e.g. are based on camera-supported testing methods such as systems based on vision sensors.
What are vision sensors and what is their field of application?
Many manufacturers use the term “vision sensor” to describe an image processing product, which can replace other types of sensors. The typical application for vision sensors is no longer only the reading of codes, but also presence-, completeness- and position-check during the production. But even that is about to change. Vision sensors are becoming more flexible and are used for more application areas. Therefore they are going to develop into a well-priced alternative to the smart camera.
The crucial characteristic of a vision sensor in comparison with a smart camera is that the user does hardly need any programming skills to get the sensor going. Therefore many maker are offering a reduced configuration possibility via drag-and-drop programming. Or else, custom-designed sensor models, from which the user can chose the sensor, which suits his application the most. Sometimes conventional vision systems are to expansive for a simple task and that is where vision sensors are in demand.
Vision sensors as well as smart cameras are machine vision systems, whose task is not only to capture an image and pass it along, but rather to interpret an image and to trigger a reaction accordingly.
Basically, all machine vision systems have the same setup. All are in possession of a camera, lens, lighting and a software for image processing and evaluation as well as a visual display unit.
When specifically comparing smart cameras with vision sensors, overall they only differ in functional range. Thus the smart camera offers a combination of different inspection possibilities, which the camera can access via a flash memory. Whereas one can say that vision sensors are specialized for certain applications. These applications are stored in the sensor via software. Examples are simple color inspections, measurements, position determination and countig pixels.
Vision sensors are by no means reduced to only one inspection task any more. With many sensors it is already possible to carry out more than one inspection task at the same time. In this connection the special feature is not the multitude of the simultaneously operable inspection tasks, but their variety.
During the production of beverage bottles, the vision sensor has to check if the cap is closed, askew or broken, if the fill level is right, the label is positioned correctly, and of course if the bottle has and scratches, cracks or is otherwise damaged.
The use of vision sensors in the beverage bottle industry
The use of vision sensors in the beverage bottle industry is manifold, no matter if empty bottle, label, fill level or cap inspection. For example during the automated return of empty bottles a vision sensor can be used to identify damaged bottles and non-brand bottles. This can be carried out by recognizing the bottles by means of several characteristic such as e.g. height, diameter, color or label of the bottles and therefore the system is able to discharge damaged bottles or bottles of a different type. This allows an early recognition already in the empties bottle case. And what is more: with an additional vision sensor for reading OCR/OCV the bottle can be identified according its number on the bottom of the bottle.
A vision sensor can also be used for identifying if the cap of the bottle is askew. But the vision sensor can do more than that. Starting with the presence check of the bottle cap and the inspection of the lock ring up to the inspection of the cap height.
But also a vision sensor proves of value in the bottling plant for checking if the fill level is matched by every bottle. An image of the bottle neck is captured and the fill level measured. Also before the bottling the bottle neck is checked for cracks or if it is splintered.
Yet another application for vision sensors is the label inspection. This requires some flexibility, as the labels not only vary along different kinds of drinks and packaging but also in language, origin and size. The inspection is carried out in free run on the conveyor band, on any kind of bottle, impartial to the type and position of the label.
Specially for the beverage bottle industry
For the inspection of beverage bottles, EVT has come up with a vision sensor especially for the beverage industry, the EyeSens BI (Bottle Inspect).
This vision sensor contains prefabricated programs to inspect bottles if the cap is closed, open, askew or missing. Also there is a program to check bottles with cork especially for wine bottles. Additionally, not only the position of the label can be checked but also if the bottles have the same label as the whole charge.
With the corresponding web interface, inspection programs can be configured via intuitive command icons directly on the object – completely without programming skills. The user not even has to install a complex software. He only has to type in the IP address of the camera into the web browser. Then the user interface opens in the browser window and the vision sensor is ready for action.
For evaluation of characteristics with an industrial machine vision system it is necessary to determine good/bad, big/small, right/wrong etc. Often the parametrization of a vision sensor can be carried out without programming skills or special training. When inspecting an object, the parametrization takes place via a configuration process. After once teaching in the parameter on a sample object the inspection can run autonomously without a PC.
Finally, one can say that the application areas for vision sensors are growing constantly and that they are more and more an alternative for the smart camera. Especially in the beverage industry. As vision sensors are becoming more flexible and are a well-priced solution.
Tel: +49721 626905-82