A statistical mask-matching approach for recognizing handwritten characters in Chinese paleography

Because most of the characters in the rare books transcribed by ancient calligraphers were contaminated by noise, to retrieve information from these books, we resort to the statistical approach instead of the structural approach. In our approach we generate one positive mask and one negative mask for each distinct class of characters in the training phase. The positive mask of a class of characters is built by finding the bits of the character images that are reliably black. Likewise, the negative mask is built by finding the bits that are reliably white. Then, we can recognize an unknown character by finding the prototype character whose masks are best fitted for the unknown character. Experimental results show that our approach performs well in this application domain.