Vision of simplicity

Keeping a high speed line on the move is the ultimate imperative of any packaging operation, and automated high performance quality inspection can be essential to avoid costly and unnecessary downtime - and sometimes punitive fines
from customers.

New developments in vision sensor technology are offering a cost-effective, but high performance solution that can keep packaging processes running. Importantly, they offer packaging engineers a simple, value-for-money and easy-to-use alternative to smart cameras.

Both vision sensors and smart cameras incorporate hardware capable of high speed inspection tasks with image processing software. But the more sophisticated hardware and the greater programming flexibility of smart cameras can bring a cost and level of complexity that are not always needed or desirable.

Vision sensors are designed to be intuitive, even for users who are not vision experts, and can be customised for applications like inspection, positioning, or code reading.

Setting up a vision sensor is not so much a programming exercise, as a teaching exercise. The engineer presents the vision sensor with examples of how the pack should look. If the pack being inspected meets the specification, then it passes. If not, an output is triggered and the pack is flagged as defective. The three most common quality issues can often be managed effectively with vision sensors.

1. Incorrect caps, missing lids, and open containers
Containers with liquids must be capped or sealed to avoid leakages that can cause production problems or contaminate content. To check for unsealed or damaged containers, a vision sensor can be taught to look at the container's size and shape.

If the container is undamaged and properly sealed, the size and shape will be within tolerances of the known 'good' container. If not, the result will trigger an output to reject the container.

The engineer can also teach the sensor to look for specific defects by presenting it with examples of those defects. To determine if the bottle cap's safety seal is intact, for example, the engineer can teach the sensor to look for a dark horizontal line at the safety seal's location. Simply present a bottle to the vision sensor with the line visible, then associate the image showing the dark line with a defect. The sensor will then flag any bottles coming down the line with this characteristic.

2. Missing or misapplied labels
To check if a label is missing or out of position, the vision sensor counts the number of pixels of the correct greyscale level in the area of interest. When looking at the position, tools can be used that rely on relative positioning - edge-finding, blob and pattern tools can be used in combination to help determine the actual pass/fail state of the output. A blob tool could be used, for example, to check for presence, and a pattern tool could verify if it is indeed the correct label.

3. Detect wrong labels
Similarly, to detect an incorrect label, a vision sensor does not need optical character recognition to tell the difference, for example, between a box of Danish pastries marked 'Apple' and one marked 'Blueberry'. Pattern matching tools allow users to check the label, to make sure it matches the type of pastry being produced.

This type of check can be very important when dealing with products that could contain allergens. Other label marking errors are similarly easy to find: missing barcodes or date codes can be detected by counting pixels in known regions of the label that are set relative to the edge tool or pattern tools, making small changes in product position less of an issue.

Vision sensors can be an excellent choice for most common packaging quality applications, leading to lower costs, higher speeds, and faster setup.

Sick launched its Inspector PIM60 vision sensor incorporating dimension verification, position detection and intelligent
measurement capability at the June 2013 Total Processing & Packaging show.

The author's views expressed on this page may not be shared by this publication.

Sick UK