Offline Handwritten Character Recognition System for Tamil Language
Abstract
Handwritten character recognition systems are gaining importance and has been done for many languages. Very few research works on handwriting recognition have been done for Tamil language. The proposed system makes an attempt to build an automatic Tamil handwritten recognition system. Building Handwriting recognition system for Tamil involves more complexities in terms of style, size and angle. Most of the existing systems use Machine Learning algorithms. However, handling huge amount of datasets(images) by Machine learning algorithms yield lower accuracy than deep learning algorithms. The proposed system uses Convolutional Neural Network(CNN) as it is widely used for processing Images and also achieves good accuracy. The Preprocessing , Feature Extraction and Classification are done by using CNN .The overall accuracy of the proposed model has obtained 92.4%.
Keywords: Convolutional Neural Network, Tamil character Recognition, Feature Extraction and Classification