Bengali Handwriting Recognition and Conversion to Editable Text

Abstract

Handwriting recognition has been a very active field of research in the past few years in different sectors. Analyzing and deducing information from various handwritings can help to tackle many ongoing issues. Though extensive work has been done for English handwritings, any progress can hardly be seen in other languages like Bengali. Hence in this paper, a system has been proposed that takes a scanned image of Bengali handwritten text as its input and after processing it gives an editable version of that text. The system consists of many phases which mainly conducts image processing, machine learning by training the neural network and lastly identification of the Bengali characters. Several data and algorithms have been used to produce a thorough and accurate result.