AUTOMATIC THRESHOLD FUNCTION FOR MACHINE VISION

In a machine vision system, a pixel-discrimination threshold level, for determining whether pixels are to be assigned one or the other of two values, is established according to parameters obtained in an initialization procedure. During initialization, a test set of items is assembled and data is entered representing indicia appearing on each item of the test set. It is then determined, for each of a range of possible threshold levels, whether the machine vision system correctly recognizes the indicia on the respective test items. Based on the thresholds which permitted correct recognition, threshold parameters are calculated for each test item and average parameters are calculated from the test item parameters. The average parameters are used to set pixel-discrimination thresholds for subsequent recognition operations.