Multifont size-resilient recognition system for Ethiopic script

  • Authors:

  • Yaregal Assabie;Josef Bigun

  • Affiliations:

  • Halmstad University, School of Information Science, Computer and Electrical Engineering, 301 18, Halmstad, Sweden;Halmstad University, School of Information Science, Computer and Electrical Engineering, 301 18, Halmstad, Sweden

  • Venue:
  • International Journal on Document Analysis and Recognition
  • Year:
  • 2007

Quantified Score

Hi-index 0.01

Visualization

Abstract

This paper presents a novel framework for recognition of Ethiopic characters using structural and syntactic techniques. Graphically complex characters are represented by the spatial relationships of less complex primitives which form a unique set of patterns for each character. The spatial relationship is represented by a special tree structure which is also used to generate string patterns of primitives. Recognition is then achieved by matching the generated string pattern against each pattern in the alphabet knowledge-base built for this purpose. The recognition system tolerates variations on the parameters of characters like font type, size and style. Direction field tensor is used as a tool to extract structural features.