摘要 |
PROBLEM TO BE SOLVED: To eliminate the need for repetitive calculation in matching without giving any influence to matched results without selecting any model character. SOLUTION: This character recognizing device is constituted in such a way that a feature point and segment extracting section 13 extracts feature points and segments through an input section 11 which inputs information from the outside and a line thinning section 12 and a labeling section 14 labels each segment with the ID of the field. Then a template generating section 15 generates templates at every field based on the labeled segments and stores the templates in a template storing section 151 and a stroke extracting section 16 labels each segment of an input character based on the information stored in the storing section 151 and extracts the segments having the same label as one stroke. In addition, a discriminating section 17 discriminates the extracted stroke for truth and falsehood based on the knowledge stored in a knowledge storing section 171 and an output section 18 outputs the information on the extracted stroke which is discriminated as 'truth'.
|