Tifinagh Printed Character Recognition through Structural Feature Extraction

1. Introduction

Optical Character Recognition system is a tool that facilitates man-machine interaction. It is a procedure that convert a text image into readable and easy to modify text by computer or similar material. Many researches has been concentrated into creating an efficient Optical Character Recognition system, especially Latina, Chinese (Mansi, 2013) and Arabic (Lorigo, 2006) characters. It is used in several areas where the work is based on the text documents, mainly in office for indexing purposes, automatic archiving of documents and banks to facilitate the reading of the amounts of checks. In the other hand, the recognition of the Amazigh characters, called Tifinagh, remains less explored.

Amazigh is a Chamito-Semitic language derived from ancient Berber. It includes a variety of dialects present from Morocco to Egypt, passing through Algeria, Tunisia, Mali, Niger and Libya. Berbers, mainly Tuaregs, use Tifinagh alphabet. It was formerly a consonantal alphabet. This alphabet has undergone modifications and variations inevitable since its origin until today. The Tifinagh characters normally written from left to right and vertically from top to bottom. Figure 1 illustrates Tifinagh characters adopted by IRCAM (Royal Institute of Amazigh Culture). It is composed of thirty-three character representing consonants and vowels. Several researches studies have been performed on the computerization of Tifinagh characters. Most of these researches are concentrated in Morocco and Algeria.

Figure 1.

Tifinagh of the Royal Institute of Amazigh Culture, IRCAM