Chain code feature based recognition of handwritten Gujarati numerals

Main Article Content

Ankit Sharma
Dr. Dipak M. Adhyaru
Dr. Tanish H. Zaveri

Abstract

This paper describes chain code based method for handwritten Gujarati numeral recognition. Literature review on Indian OCR indicates that in comparison with Bangla, Hindi, Kannada, Tamil and Telugu scripts, the OCR activities related to Gujarati script is very less. Development of OCR for Gujarati script is quite challenging area for research. In this work, recognition of isolated Gujarati handwritten numerals is performed using chain code based methods. Horizontal scanning and maximum distance from centroid methods are used for deciding the starting point for calculating the chain code sequence. An overall accuracy of 96.37% and 95.62% is obtained using feed forward neural network classifier by the proposed methods respectively. One of the significant contributions of this paper is towards the generation of large and representative database for handwritten Gujarati numerals.

Keywords:Chain code, Gujarati handwritten numeral recognition, Neural network classifier.

Downloads

Download data is not yet available.

Article Details

Section
Articles