PURPOSE: To decode one-stroke alphabets, numbers, etc., at a high recognition rate by extracting features which show position information on the terminal point of a character or symbol and its direction, features which show the relation between the terminal point and a character frame, features which show a projection on XY coordinates, and features determined by the projection.
CONSTITUTION: A terminal point and character frame information extractor 4 calculates the coordinates of the starting point and ending point of a character from a coordinate point sequence storage part 1 by referring to information on a character frame from a character frame information storage part 3, and outputs the coordinates to an extracted data storage part 7. A terminal point information extractor 5, on the other hand, refers to the coordinates of the terminal points from the storage part 1, and a projection information extractor 6 finds projections on the X coordinate the Y coordinate from data of the storage part 1, and refers to the Y coordinate of the center of the character frame of the storage part 3 for arithmetic to output the result to the storage part 7. Then, a recognition part 9 makes comparisons with respective one- stroke capital alphabets, numbers, special symbols, etc., filed previously in a dictionary storage part 8 to output the most similar features.
YOSHIDA MASUMI
KATOU OSAMU
SHIMIZU SHINICHI