Devanagari Text and Calligraphy Recognition Using ICF & ACF

Abstract

In India Multilanguage is used for communication. On average, 600 million people use the Devanagari script for documentation. As it is a national script; hence, the information & communication technology era gives more scope to look at OCR strategies. Recognition of handwritten language text has been a demanding task in the computing field. This paper provides an adept approach for recognizing printed and handwritten documents for English and Devanagari Script. The major contributions in this paper are
1. Preparation of the datasets for handwritten Devanagari compound characters, words and calligraphy text
2. To provide an adept approach for recognizing English and Devnagari text
3. Analysis of pattern recognition methods based on Wavelet, Contour and SVM
4. Detailed discussion on results obtained for several text categories like Devanagari compound, alphanumeric, confusing, mirror characters, and calligraphy.
The present work consists of wavelet transform, statistical parameters Inter correlation and autocorrelation functions, support vector machine, and contour. The novelty of work is recognition of text and calligraphy using Correlation functions -ICF, ACF and template generation for new patterns. . Standard dataset HPL (Devnagari characters and numbers) and own dataset (2146 – mentioned categories) are used for experimentation. The recognition accuracy achieved for printed text is 100%, handwritten text is 98.89 and devanagari compound character is 98.87. This throughput shows superior performance than existing work. Handwritten character recognition is useful solution for society for day to day activity.