Peripheral Contour Feature Based On-line Handwritten Uyghur Character Recognition
Mayire Ibrayim, Dilmurat Tursun, Askar Hamdulla, Zulpiya Kahar · 2013
In this paper, we conduct a deep research on the grid direction feature, peripheral contour feature, stroke number feature and additional part’s location feature for recognition of 32 independent forms of Uyghur characters through fully considering the unique shapes and writing styles of Uyghur characters. We divide the whole characters to subclasses by applying stroke number feature and additional part’s location feature of individual characters, and then conducted several recognition experiments using the above features alone and various combinations. The experimental results show that peripheral contour feature is the best choice. The design of the character recognition in this article is an important part of the string recognition / word recognition, and laid the foundation for the implementation of Uyghur character handwritten input method in various platforms.