Handwritten Malayalam Character Recognition using Regional Zone with Structural Features

Optical Character Recognition (OCR) extracts features from an image of script and converts it to machine-readable code. OCR comprises of line segmentation, word segmentation, character segmentation and character Recognition. Printed documents are efficiently converted to the editable text format with almost 100% accuracy. Handwritten character recognition places difficulties in identifying and translating scripts because of the wide variation in human handwriting. The writing styles like line spacing, word spacing, character sizes and shape of each character varies from person to person. Feature extraction and character recognition are different for different languages and become the most complicated task among the phases of OCR. By language characteristics, feature extraction can differ for each language. The Malayalam characters are characterized by their curved and non-cursive nature. The handwritten character recognition for the Malayalam language that proposed here uses a regional zone based method with structural feature extraction.