Web inspection using gradient-indexed optics

Abstract

Photoelectric sensors, used to inspect product and packaging attributes, are ubiquitous in manufacturing processes of today. Most common are single element sensors that detect the presence or absence of an attribute, product, or package as a binary on/off signal. In the Web converting industry, more detailed information is often needed of the Web than a single sensor can detect. In addition, single element sensors are often not sufficient to reliably sense product attributes. A small amount of contamination can render single element sensors dysfunctional. They are also hindered from being able to distinguish between components on products that respond similarly to the light source but may differ in geometric shape on the product. Machine vision systems can detect much more detail and have proven effective for many challenging applications. However, due to their complexity, they are generally specified only for the most challenging or most critical quality inspections. This paper will present a new, hybrid approach to product attribute sensing employing desirable aspects of both single element sensors and machine vision systems without many of the disadvantages of each. The hybrid sensor contains a linear array of pixel elements along with a gradient-indexed lens array. The linear array allows for greater sampling and geometric characterization in one dimension while the lens array allows for very close standoff distances from the web and a 1:1 correspondence of object to image. The presentation of this sensor technology includes a tutorial on the key technical properties of gradient-indexed optics and experimental results on a composite Web running on a trial machine