An MLP based Approach for Recognition of Handwritten `Bangla' Numerals

TL;DR

This paper presents an MLP-based pattern classifier for recognizing handwritten Bangla digits, achieving high accuracy and potential applications in OCR systems for Bangla script recognition.

Contribution

The paper introduces a novel feature set and an MLP classifier specifically designed for handwritten Bangla digit recognition, demonstrating high accuracy on a sizable dataset.

Findings

Recognition accuracy of 96.67% on 6000 samples

Effective feature set including shadow, centroid, and longest-run features

Potential for extending to full Bangla character OCR

Abstract

The work presented here involves the design of a Multi Layer Perceptron (MLP) based pattern classifier for recognition of handwritten Bangla digits using a 76 element feature vector. Bangla is the second most popular script and language in the Indian subcontinent and the fifth most popular language in the world. The feature set developed for representing handwritten Bangla numerals here includes 24 shadow features, 16 centroid features and 36 longest-run features. On experimentation with a database of 6000 samples, the technique yields an average recognition rate of 96.67% evaluated after three-fold cross validation of results. It is useful for applications related to OCR of handwritten Bangla Digit and can also be extended to include OCR of handwritten characters of Bangla alphabet.