METHOD AND DEVICE FOR ROUGH CLASSIFYING INPUT CHARACTERS IN ON-LINE CHARACTER RECOGNITION

摘要

PROBLEM TO BE SOLVED: To perform an efficient rough character classification with a simple configuration also dealing with one-stroke characters and to roughly classify even the characters, which has not be able to be roughly classified by a conventional method, by extracting feature points from an input character, generating a bit map connecting these feature points through straight lines and performing the rough classification by a feature vector extracted from the bit map. ;SOLUTION: First of all, the coordinate data sequence of the input character containing off-stroke information provided by coordinate data sequence generating processing 11 is normalized in the direction of factors of vectors with respectively independent values by normalizing processing 12. Next, the feature points are extracted by feature point extracting processing 13. Afterwards, a straight line for connection the feature points is generated by bit map preparing processing 14. In feature vector calculating processing 15, a feature vector is calculated from the bit map and in rough classifying processing 16, the feature vector is compared with values stored in a rough classification dictionary 17 so that rough character classification can be performed.;COPYRIGHT: (C)1997,JPO