Identification of submil-defects on crystal blanks with digital image processing technique
Quartz resonators play an important role in satellite communication as well as in the time keeping of our daily life. Detection and identification of the extraordinarily fine scratches and cracks on crystal blanks by computer not only help relieve a human operator from carrying out this screening job with the naked eye, but also help assure quality and reliability, increase productivity and reduce the per-unit cost. Difficulties encountered in these detections are due to the fact that such defects are almost indiscernible and indistinguishable from each other when thin. The purpose of this thesis is to develop an efficient algorithm for the analysis, detection and identification of various types of these defects on real quartz crystal blanks. Experimental results show that tiny scratches and cracks were successfully detected, and identified with the algorithms and methods.