Title :
Shape inspection of crystal glassware
Author_Institution :
Southampton Univ., UK
Abstract :
Video rate systems have been developed aimed at improving the inspection of quality crystal glass. One system was aimed at providing an operator with a navigational aid to comparison. This presented the outline of the glass currently inspected, together with the outline of a master glass, for an operator to compare the two. A simple algorithm was developed which was then translated to a hardware implementation so that video rate inspection could be achieved at low cost. The system was however limited by the need for close attention of the operator. A preliminary stage in automatic analysis concerns the analysis of the best match between the images when both difference in position and size have been compensated for. A least squares algorithm has been developed to produce the match which is an efficient algorithm which lends itself to a hardware implementation. The algorithm evaluates the scale factor and offset to achieve the best match between a current outline and a master outline. This is an important precursor to analysing the match between the glasses. However the least squares match between the glasses appears, by itself, insufficient for automatic analysis alone and research is in progress to capitalise on this technique. The technique lends itself to a hardware implementation for fast registration of images and considerations for possible hardware implementation have been included
Keywords :
computer vision; inspection; least squares approximations; picture processing; automatic analysis; computer vision; crystal glassware; fast registration of images; hardware implementation; least squares algorithm; navigational aid; shape inspection; video rate system;
Conference_Titel :
Image Processing and its Applications, 1989., Third International Conference on
Conference_Location :
Warwick