The most economical way to inspect optical components for imperfections has been found to be by using the eye to collect the low-angle light they scatter. Attempts to automate this process have so far not succeeded due to considerations of sensitivity and cost. The use of the eye, however, becomes impractical when the surface to be inspected is large and moving rapidly, as is the case with continuous-strip product inspection.
Optical quality, including appearance, is of great importance to manufacturers of most strip products. Automotive glass manufacturers, particularly those making glass for laminated windscreens, where driver viewing at steep angles of incidence occurs, find this configuration magnifies the effect of very small surface irregularities. Moreover, the need to detect a particle of diameter 0.1 mm across a strip of float glass 4 m in width at a rate of 1000 particles per second, should they occur, presents an extremely demanding problem. Although CCD cameras have been tried, the most satisfactory solution found so far, with capability of scanning up to three times faster, involves the use of laser beam scanning.
The range of imperfections likely to occur with float glass includes seeds, bubbles, stones, tin oxide, and tin, while contaminants include, for example, glass dust, flakes, chips, roll marks, and insects. Thresholding and classification based on measurement and experience is vital to ensure maximum product yield. Contaminants, which are removable, may arise with an incidence of 1000 times that of imperfections, which are not removable.
Since the task of automatic inspection is likely to be expensive, it is important to understand how it can be justified. Customers for the product will wish to pay a price related to a quality grade adequate for their application. Accurate quality grading can only occur when objective quality measurement is possible. Under these circumstances, overspecification can be avoided and productivity increased. Objective measurement, supported where possible by traceability to national standards and involving ISO 9000, should lead to greater customer satisfaction. Finally, the application of modern technology, avoiding the need for the human inspector, should enable faster and more reliable quality grading. These factors and others are illustrated in Fig. 9.1.
Online access to SPIE eBooks is limited to subscribing institutions.