Printed Chinese Character Recognition Based on Pixel Distribution Probability of Character Image

Author

Wang, Ning

Author_Institution

P. R. China Sch. of Comput. Sci. & Eng., South China Univ. of Technol., Guangzhou

fYear

2008

fDate

15-17 Aug. 2008

Firstpage

1403

Lastpage

1407

Abstract

This paper analyzed the disadvantage of the character recognition method of Casey and Nagy. It adopted advantage of two methods and proposed a new method of character recognition. In this method, the pixel distribution probability of character image is the basal index. The minimal distance product and the maximal fuzzy correlation measure are the evaluation index. It not only overcame the shortcoming of above two methods but also improved the rate of character recognition. With the recognition experiment of 3500 printed Chinese characters, the result showed that this method is accurate and effective since the recognition rate of single character exceeded 99.97%.

Keywords

character recognition; feature extraction; fuzzy set theory; matrix algebra; probability; character image; maximal fuzzy correlation; minimal distance product; pixel distribution probability; printed Chinese character recognition; Bismuth; Character recognition; Computer science; Computer security; Feature extraction; Image segmentation; Information security; Probability; Signal processing; Smart pixels; Pixel Distribution Probability; Printed Chinese Character; Recognition; the maximal fuzzy correlation measure; the minimal distance product;

fLanguage

English

Publisher

ieee

Conference_Titel

Intelligent Information Hiding and Multimedia Signal Processing, 2008. IIHMSP '08 International Conference on

Conference_Location

Harbin

Print_ISBN

978-0-7695-3278-3

Type

conf

DOI

10.1109/IIH-MSP.2008.118

Filename

4604304