Handwritten Devanagari Character Recognition using Neural Network

Abstract

In this digital era, most important thing is to deal with digital documents, organizations using handwritten documents for storing their information can use handwritten character recognition to convert this information into digital. Handwritten Devanagari characters are more difficult for recognition due to presence of header line, conjunct characters and similarity in shapes of multiple characters. This paper deals with development of grid based method which is combination of image centroid zone and zone centroid zone of individual character or numerical image. In feature extraction using grid or zone based approach individual character or numerical image is divided into n equal sized grids or zones then average distance of all pixels with respect to image centroid or grid centroid is computed. In combination of image centroid and zone centroid approach it computes average distance of all pixels present in each grid with respect to image centroid as well as zone centroid which gives feature vector of size 2xn features. This feature vector is presented to feed forward neural network for recognition. Complete process of Devanagari character recognition works in stages as document preprocessing, segmentation, feature extraction using grid based approach followed by recognition using feed forward neural network.