A Machine Intelligence Based Model for the Classification of Odia Printed and Handwritten Images

Abstract

Language plays an important role for the communication among all of us. It is very much essential to detect the printed and handwritten language from several images to extract crucial information from it. In this paper, a machine intelligence (MI) based model is proposed for the classification of Odia printed and handwritten language (OPHL) from the analysis of several Odia language images. The proposed approach is mainly focused on the machine learning (ML) based hybridization mechanism. This mechanism focuses on the combination of the ML based methods such as Logistic Regression (LR) and Neural Network (NN) for the classification of printed and handwritten Odia images. The proposed method is compared with the ML based methods such as Support Vector Machine (SVM), Decision Tree (DT), K-Nearest Neighbour (KNN), AdaBoost (ADB) and Random Forest (RF) methods in terms of
Classification Accuracy (CA) performance metric. The proposed method is able to classify the Odia printed and handwritten images in a better way as compared to other methods. The simulation of this work is carried out using Orange 3.26.0.