Fool-proof approach to quality

FOR the six operators who are lined up along the production line of a soft drink bottling factory, their instructions are straightforward: ensure that the right caps are correctly placed on bottles,and that manufacturing and expiry dates are printed on each product.

Whilst this may appear to be a simple quality control task, the tricky part is in ensuring that every bottle that passes by the operators is thoroughly checked for the inconsistencies and errors. At production speeds that could range from 30,000 to 50,000 bottles per– read rates of 93-95 percent to over 99 percent can be achieved.

The tool is based on image processing algorithms and comes in an easy-to-use programming interface. First, the segmentation process breaks down each line, sets regions around each of the characters and automatically segments them. In the next step, segmentation parameters are adjusted to place a box around each character.

Each character is broken down further into small fragments and the Segmenter finds each letter based on characteristics such as skew, min/max height, min/max width, minimum aspect ratios, angle/skew, intercharacter gap and intracharacter gap. These and other characteristics define the font for OCRMax. Finally the characters are trained so they can be recognised by the system. 

Users can train an entire string of characters in one single step. They can also train or remove single characters, train multiple instances of same characters, and load or save trained fonts to an application. The Segmentation tool also offers a noise filter, stroke width filter, compensation for changing lighting conditions, and automatic character scaling. These features all help to improve read rates.

Assurance of quality
Besides recognising text, the vision system can be configured to prevent the double-labelling of bottles. Furthermore, cameras can be set up to check for the presence of any foreign objects in the bottles. Having such a system in place also  means that the entire operation becomes less labour intensive. In this instance, the number of operators required  on  each  shift is reduced from six – to just one. This operator can monitor the production line in real-time, via a user-friendly interface.

When errors are discovered, the system pauses the production line and alerts the operator. He can then view captured images of the product to determine the cause/s of the problem. Alternatively, the vision system can be configured to divert defective products into a rejection bin for collection after the work shift.

Machine vision offers a wide range of QC options, customised to individual applications. By adopting this technology, a labour intensive facility can be quickly transformed into an efficient, error-free production environment.