An approach for automatic Indic script identification from handwritten document images

Abstract

Script identification from document images has received considerable attention from the researchers since couple of years. In this paper, an approach for HSI (Handwritten Script Identification) from document images written by any one of the eight Indic scripts is proposed. A dataset of 782 Line-level handwritten document images are collected with almost equal distribution of each script type. The average Eight-script and Bi-script identification rate has been found to be 95.7% and 98.51%, respectively.