A Study on Writer Identification using a Pile of Signatures

Abstract

In this paper, We suggest a method of writer identification using a pile of signatures. In the method, after each character of the individual signature is separated, the size of the character is normalized and its center of gravity is calculated to pile up characters. From experimental results using 18 writers' signature, it is found that the average identification rate is 94.1% without normalization and 98.7% with normalization.