Handwritten Text Recognition Using Information Energy And Neural Network

Kanchan Keisham, Sunanda Dixit

Abstract: In handwritten character recognition one of the most challenging task is segmentation. This is mainly due to different challenges like skewness of textlines, overlapping characters, connected components etc. This paper proposes a character recognition method of handwritten English documents. The textlines are segmented based on information energy that is calculated for every pixel in the scanned document and the characters are recognized using artificial neural network (ANN).

Keywords: line segmentation, word segmentation, character segmentation, information energy, neural networks, recognition etc