Optical Bangla character recognition using chain-code

This paper presents an optical Bangla character recognition (OCR) system using Freeman Chain Code and feed forward back propagation neural network which provides much higher performance than some other traditional approaches. After scanned the printed Bangla document, preprocessing and segmentation has done. In the feature extraction stage Freeman Chain Code is used. Finally, feed forward back propagation neural network is used in classification of Bangla character. Performances results are presented in the literature.

I. Introduction

Bangla is one of the most widely spoken languages in the world. About 250 million people speak in this language all over the world including Bangladesh and West-Bangle. Optical Character Recognition (OCR) is an important branch of Pattern Recognition. For some scripts languages like Bangla researches are not mature enough in this field.