Spot checks on quality control

Camera technology developed at the Fraunhofer Institute for Industrial Mathematics could be used to eliminate bottlenecks in paper, textile and wood production production.

Camera technology that is capable of spotting tiny defects moving past at high speed could be used to eliminate bottlenecks in paper, textile and wood production.

Though modern paper factories can produce enough paper to cover a football pitch in less than 20 seconds, the quality of the paper usually has to be checked manually, creating a delay in the production process.

So scientists at Germany’s Fraunhofer Institute for Industrial Mathematics have developed a system of cameras and image-analysis algorithms to enable automatic high-speed checking.

According to the Institute’s IT specialist Markus Rauhut the software is key. ‘The hardware components are only one aspect of the system. The attainable speed and precision of quality control depend, above all, on the algorithms of the image processor.’

Known as Spot, the system identifies imperfections in paper such as glossy patches, scratches, perforations and indentations less than 1mm in size as they travel by at up to 100kph.

The data that the system gathers is fed back directly to enable real-time changes to the production process. Spot can be operated using standard PCs and expanded modularly as required.

Because of the vast amount of paper passing by, analysis of every byte of data from every image would overwhelm the software. So various electronic filters are employed to extract identifiable ‘typical’ flaws, which are then reproduced in a new image as areas of interest. What constitutes a ‘typical’ flaw is defined before the production run begins.

The system could also be used to monitor textiles and wood production. But what works well with uniformly coloured paper would require much greater computational performance for a wider range of materials, said Rauhut.

‘If flaws in patterns are to be optically analysed at high speed, it is essential that the existing algorithms be further simplified. Our systems would be suitable for quality control in these areas – but they have to be ‘trained’ first. Then they will surpass the speed and precision of the human eye with no problems.’