Machine vision using image data feedback for fault detection in complex deformable webs

Complex deformable webs like lace fabrics are amongst the most complicated materials to be inspected without any human intervention. The ability of such webs to deform easily, along with their complex patterns, presents a challenging problem for any machine vision system. Their pattern complexity makes it very difficult to use any global image processing or rule-based techniques for inspection, leaving direct comparison with a perfect prototype as the only choice available. Due to the elastic nature of the lace fabric, small distortions are characteristic of the product and are unavoidable. This renders sensitive direct comparison methods, such as those employed for the automatic inspection of printed circuit boards, prone to unacceptable false alarms. In the present work, the authors have exploited a mechatronic methodology, coupled with image processing techniques, to develop a high-speed inspection station for complex patterned lace webs.