| 摘要 |
PROBLEM TO BE SOLVED: To solve the problem that, in a conventional method, input of characters is troublesome due to the necessary to input voiced sound symbols and semi-voiced sound symbols in a small entry frame for voiced sound symbols and semi-voiced sound symbols, and that, in another method, recognizing performance is deteriorated due to the impossibility to correctly extract features when even one of conditions for judging the voiced sound symbols or the semi-voiced sound symbols is not unsatisfactory. SOLUTION: In this character recognizing method, the coordinate point information on handwritten characters described in a character frame having a voiced sound symbol and semi-voiced sound symbol are is obtained from an inputted character pattern, the voiced sound symbol and semi-voiced sound symbol information on the input pattern is detected by using the input coordinate point information and the voiced sound symbol and semi-voiced sound symbol area, and the classification of a feature dictionary in which the feature values of characters to be recognized are preliminarily stored by classifying the feature values into characters in which no voiced sound symbol or semi-voiced sound symbol strokes are included is selected by using the voiced sound symbol and semi-voices sound symbol information, and the feature values of the characters to be recognized are collated with the feature values of the input pattern by a dictionary collating means, and the recognized result is obtained.
|
