Combining Zernike moments with Regional features for classification of handwritten ancient Tamil scripts using Extreme Learning Machine

Abstract

Handwritten Tamil character recognition is one of the active areas in research. Due to high variability of writing styles, developing handwritten character recognition system is a big challenge. The concept proposed gives a way to perform classification of handwritten ancient scripts in Tamil, which is one of the oldest languages in India. The approach utilizes Extreme Learning Machine for classification of handwritten ancient Tamil scripts. The Extreme Learning Machine is trained by Zernike moments and Regional features. The performance of Extreme Learning Machine is compared with Probabilistic Neural Networks. From the experimental results it is inferred that Extreme Learning Machine gives highest accuracy rate of 95%.