Recognition of printed Telugu characters

Abstract

A two-stage recognition system for the recognition of the Telugu alphabet (with more than 2000 different characters) is presented. It is based on the observation that the Telugu characters are composed of certain primitive shapes and basic letters. In the first stage of the recognition process, a knowledge-based search is made to recognize and remove the primitive shapes that are present in the input character. A directed curve-tracing method is used for this purpose. In the second stage, the pattern obtained after the removal of primitives is coded by tracing along points on it, and the code thus obtained is used to identify the basic letter. On the basis of the knowledge of the primitives and basic letter present in the input pattern, classification is achieved by means of a decision tree. The recognition scheme has been tested on hand-digitized and computer-generated characters. Computer simulation results are given.