Abstract: The present invention is directed to a method for recognizing and classifying one or more unknown patterns, such as letters or numerals, by comparing these unknown patterns to a set of reference patterns. The unknown patterns, as well as the reference patterns, are irradiated by a coherent or incoherent beam, and correlation outputs are measured which would be at peak value of an auto-correlation output or a cross-correlation output between the unknown patterns and the reference patterns. Membership functions are created for each of the reference patterns, and a membership value is estimated or calculated on the basis of the output correlation between the unknown patterns and the reference patterns. Based upon this membership value, the specific unknown pattern is determined.