A method for the recognition of Japanese hiragana characters

A method for the recognition of Japanese hiragana characters is described. To recognize Japanese hiragana characters having complicated outlines, rotation indices are introduced. Sequences of the rotation indices contain information of the figure of the character. Smoothings and irregularity processing operations are performed for an economy of memory. The method developed will be applicable to recognition of other characters. The experimental results were obtained by simulation on a FACOM 231 computer.