Earticle

초록

영어

In this paper, we propose a word recognition method for printed Arabic word images using HMM and Levenshtein Distance. The existing algorithm has the difficulty for Arabic text recognition to treat various fonts and sizes. This is because Arabic characters are cursive and each character may have up to four different shapes based on its location in a word. Our work begins with segmentation of a word into characters. Then each character is recognized individually using HMM classifier. Since the recognition of HMM is not accurate enough, we apply Levenshtein distance to correct misclassification and miss segmentation of a character in a word. Levenshtein distance works by comparing between recognized word and every words in a dictionary. We tested our proposed system with APTI dataset, and the achieved average recognition rates in more than 95% for six different fonts.

목차

Abstract
 1. Introduction
 2. Proposed Approach
  2.1. Preprocessing
  2.2. Segmentation
  2.3. Feature Extraction
  2.4. Classification
  2.5. Post Processing
 3. Experimental Results
 4. Conclusion
 Acknowledgements
 References

키워드

저자

• Perdana Adhitama [ School of Electronics and Computer Engineering, Chonnam National University Gwangju, 500-757, South Korea ]
• Soo Hyung Kim [ School of Electronics and Computer Engineering, Chonnam National University Gwangju, 500-757, South Korea ]
• In Seop Na [ School of Electronics and Computer Engineering, Chonnam National University Gwangju, 500-757, South Korea ] Corresponding author

참고문헌

발행기관

간행물

표지보기 관심저널 등록

• 간행물명

  International Journal of Software Engineering and Its Applications

• 간기

  월간

• pISSN

  1738-9984

• 수록기간

  2008~2016

• 등재여부

  SCOPUS

• 십진분류

  KDC 505 DDC 605