Affiliation:
1. Lanzhou University of Technology
Abstract
A method based on Bernoulli mixtures is presented in this paper for the purpose of describing the pattern of probability distribution for each of the data sets of handwritten numerals (0-9), and then classifiers are formed from these. The test samples are recognized by their posterior probabilities conditioned on every classifier. Experimental results show that our method is superior to conventional methods on robustness and accuracy.
Publisher
Trans Tech Publications, Ltd.
Reference8 articles.
1. Y. Al-Ohali, M. Cheriet, and C. Suen, Databases for recognition of handwritten Arabic cheques, Pattern Recognition, 36: 111–121, (2003).
2. A. Juan and E. Vidal, On the use of Bernoulli mixture models for text classification, Pattern Recognition, 35(12): 2705– 2710, Dec. (2002).
3. Hu. J and Yan. H, Structural primitive extraction and coding for handwritten numeral recognition , Pattern Recognition, 1998, 31 (5): 493-509.
4. G. Henkelman, G. Johannesson and H. Jónsson, in: Theoretical Methods in Condencsed Phase Chemistry, edited by S.D. Schwartz, volume 5 of Progress in Theoretical Chemistry and Physics, chapter, 10, Kluwer Academic Publishers (2000).
5. R. O. Duda and P. E. Hart, Pattern Classification and Scene Analysis, Wiley, (1973).