Printed character preclassification based on word structure

摘要

In today's office environment, electronic representation of information, intended as digital representation of a document, has been raised to the part of a protagonist. This paper advances a mixed topological/statistical approach to printed character preclassification, a way of separating a character set into disjointed categories. What is suggested is strictly related to the examination of particular character aggregates, for extracting special features in order to establish the membership of a character to one of seven categories. An improvement of correct recognition rate can be obtained by means of such an approach to optical character recognition.