Automated defect classification system
AssigneeInternational Business Machines Corporation
InventorsVirginia H. Brecher, Paul B.-L, Chou, Robert W. Hall, Debra M. Parisi, Ravishankar Rao, Stuart L. Riley, Martin C. Sturzenbecker
Life of this patent
Estimating the term…
Abstract
An integrated visual defect detection and classification system. The invention includes adaptive defect detection and image labeling, defect feature measures, and a knowledge based inference shell/engine for classification based on fuzzy logic. The combination of these elements comprises a method and system for providing detection and analysis of product defects in many application domains, such as semiconductor and electronic packaging manufacturing.
Claims52
A digital computer for automatically classifying defects in an object comprising:
- means for forming a defect image from a digital input image of an object;
- means for forming an adaptively labeled defect image from said digital input image and a reference label image of aid object;
- means for creating a plurality of defect feature measurements of a defect from at least said defect image, and said adaptively labeled image;
- fuzzy logic inference engine means having a plurality of if-then defect class rules each based on a plurality of fuzzy sets of fuzzy logic parameters, means for converting said defect feature measurements into fuzzy logic parameters and applying said converted parameters to said if-then rules to compute a delete class fuzzy set for each defect;
- and
- means lot deriving output data from each said defect class fuzzy set which classifies each defect.
A method for automatically classifying defects in an object comprising the steps of:
- forming a defect image from a digital input image of the object;
- forming an adaptively labeled defect image from said digital input image and a reference label image;
- creating a plurality of defect feature measurements of a defect from at least said defect image, and said adaptively labeled image;
- converting said defect feature measurements into fuzzy logic parameters;
- applying said converted parameters to a plurality of if-then rules each based on a plurality of fuzzy sets of fuzzy logic parameters;
- computing a defect class fuzzy set for each defect;
- and
- deriving output data from each of said defect class fuzzy sets for classifying each defect.
Dependencies are inferred from the claim text.