A NOVEL APPROACH TOWARDS ONLINE DEVNAGARI HANDWRITTEN WORD RECOGNITION BASED ON ROBUST FEATURE EXTRACTION METHOD AND FFNN CLASSIFIER.

In recent time the use of handwritten recognition tools are increasing in various fields of applications, therefore it becomes interesting area of research to many researchers. Different methods are presented with goal of more accuracy for character recognition with less time for different languages like English [Major], Hindi, Telagu, Gurumukhi, Kannada etc. Devnagari is also one of the most widely used scripts in India. This paper is focused to present accurate, robust and faster method for Devnagari handwritten recognition for Marathi language using efficient algorithms in each phase of handwritten recognition. There are three main phases of any handwritten character recognition methods such as Segmentation, Feature extraction and classification. In first phase, input image is preprocessed using Gaussian filter for smoothing and noise removal. Further using thresholding preprocessed image is segmented with additional morphological operations such as dilation, filling, erosion in order to get finalized segmented image. In second phase, hybrid efficient and optimized feature vector of length 91 is presented by using combination of geometrical features, regional features, distance transform and gradient features. In third phase, use of efficient, accurate classifier called Feed Forward Neural Network [FFNN] for online handwritten character recognition is presented. Also the performance of both SVM & K-NN classifiers is computed for same database. Here 200 commonly used handwritten words are collected from 50 users with different handwriting styles to create database of 10,000 words. For experimentation, 7500 word samples are used to create 20 dataset out of which 70% samples are used for training, 20% for testing & 10% for validation. Overall recognition accuracy obtained using SVM, K-NN & FFNN classifiers is 84.70%, 82.30% & 94.57% respectively.