Abstract: An optical character recognition system wherein both feature analysis and matrix analysis operations are performed to insure proper identification of a character under scan. Operating speed and reliability are enhanced by initially classifying the character under scan as being included in a first sub-group of characters having features in common with the character being scanned whereafter final indentification of the unknown character from among the selected sub-group is effected by means of a comparison operation based on a weighted matrix analysis. Preprocessing steps are taken to precisely frame the informational content of the character under scan. The preprocessing steps are effected as the information defining the character is being digitized. Simultaneously therewith the digital information is temporarily stored and, upon completion of the character framing operation, the temporarily stored information is gated into a random access memory (RAM) for use in the further processing operations.