ONLINE TELUGU HANDWRITTEN CHARACTER RECOGNITION USING EFFICIENT MACHINE LEARNING APPROACHES

Abstract

In Pattern recognition, Online Telugu Handwriting Recognition (OTHWR) has become one of the
recent research areas of interest due to exponential use of resources such as paper documents,
photographs, Smartphone, and iPods. Telugu Handwriting differs from person to person and it is
a difficult task to recognize the Telugu characters. Telugu Handwriting Recognition is
categorized into two ways - online and offline. Online Telugu character recognition involves
conversion of digital pen-tip movements into a list of coordinates. Telugu character recognition
is considered as one of the most critical components which enable a data processor to distinguish
letters and digits possibly using the contextual data. Various attempts in resolving this problem
by using different selections of classifiers and features have been established and still the
problem is remaining challenging. In the proposed work, we have used Optical Character
Recognition System and various machine learning technique i.e., Convolution Neural Network
(CNN) and Support vector machines. In SVM, we have constructed the stroke recognition engine
and the characters have been represented as a sequence of strokes and features have been
extracted and classified. Support vector machines for Telugu language (south Indian) character
recognition algorithm with high recognition accuracy and minimum training and classification of
time. A qualified analysis has performed to test the efficiency of the proposed models against
previous methods on an interesting dataset. In observations, it was found to have improved than
that of some of the recent expressions made in literature usage for the identification of online
handwritten Telugu handwriting characters.