Morphological segmentation of markings for inspection of IC packages under complex backgrounds

In the process of packaging integrated circuits (ICs), special markings are drawn to indicate the defective chips detected by test & inspection equipments. During the automated optical inspection process, these markings are detected and identified by the inspection equipment for a special treatment of detective IC packages. The marking is sometimes difficult to segment by the imaging device due to the complex background features among which the marking is drawn, such as circuit features and other graphic features. In this paper, we develop several marking segmentation algorithms based on grayscale hit-or-miss morphological transformation operations. The markings are assumed of known width different from other background features and of detectable contrast from background. Performance of the newly developed algorithms are extensively tested and compared with simple well-known morphological algorithms and the result shows equal or better segmentation accuracy.