Method for localizing and isolating an errant process step

A method for process step defect identification comprising the steps of:

• characterizing anomalies in a product said anomalies detected by a content-based imaging system, wherein the characterizing step comprises the steps of detecting a product anomaly passing within imaging range of a computer vision system;
• forming an image of said product anomaly;
• assigning defect characterization data to said formed image;
• and, storing said formed image and said assigned image defect characterization in a defect characterization database, whereby each formed image and assigned image defect characterization in said defect characterization database can be retrieved for subsequent comparison to a query image depicting a product anomaly;
• acquiring a query image of a product defect;
• correlating a particular characterized anomaly with said query image, wherein the correlating step comprises the steps of retrieving from said defect characterization database, a selection of images, each image having image content similar to image content extracted from said query image;
• ranking said selection of images according to a similarity metric wherein the ranking step comprises the steps of:
• retrieving from said defect characterization database, defect characterization data corresponding to said selection of images;
• deriving from said defect characterization data a conditional probability distribution of said defect having occurred in a particular process step;
• and, identifying a process step as a highest probable source of said defect according to said derived conditional probability distribution, and,
• identifying an errant process step associated with said correlated image.

A computer apparatus programmed with a routine set of instructions stored in a fixed medium comprising:

• means for retrieving from a defect image database a selection of images, each image having image content similar to image content extracted from a query image depicting a defect, each said image in said selection having corresponding defect characterization data;
• means for directly deriving from said defect characterization data a conditional probability distribution of said defect having occurred in at least one process step;
• and,
• means for identifying a process step as a highest probable source of said defect based on said derived conditional probability distribution.