PURPOSE: To improve the segmenting accuracy of KANA (Japanese syllabary) separation characters.
CONSTITUTION: Strokes inputted from a coordinate input means 1 are coded by a stroke coding means 2 and a basic segment dividing means 3 detects overlaps among the respective strokes, connects the strokes provided with the overlap and turns them to one basic segment. A first kind KANA separation character connection means 4 detects first KANA separation characters from a basic segment string and connects them and a character segmenting means 5 connects the basic segment strirag based on conditions set beforehand. A second kind KANA character connection means 6 detects and connects second KANA separation characters with the basic segments not connected in the first kind KANA separation character connection means 4 and the character segmenting means 5 as objects.
MATSUMOTO KOJI
ICHII HIDESATO