Print and Surface Inspection

Surface defects are often too variable to define with rules: natural texture, pattern and color differences can be mistaken for defects. That is why we combine product-specific lighting with texture analysis and deep learning in surface inspection; scratches, dents, bubbles and texture irregularities are brought out in the image and classified.

In print inspection every print is compared with a reference image, and color shifts, missing print, misregistration and stains are detected. Area or line scan cameras are used depending on line speed, and a faulty product is pushed into a reject bin by a piston or diverted to the reject line.

Some of our applications on the shop floor: deep-learning inspection of automotive headrests from several angles in a camera tunnel, dent and tear detection on sugar boxes, and finding fabric defects (oil marks, weft/warp misalignment, tears) and flaws in leather texture. The image on this page is the operator screen of our headrest station.