Multiple Classifier System for Offline Malayalam Character Recognition

Abstract

This paper presents a multiple classifier system for the recognition of offline handwritten Malayalam characters. The features used are the gradient and density based features. These feature sets are fed as input to two feedforward neural networks. The results of both these neural networks are combined using four different combination schemes: Max rule, Sum rule, Product rule and Borda count method. The best combination ensemble with an accuracy of 81.82% is obtained by using the Product rule combination scheme.

Review of offline handwritten text recognition in south Indian languages

“…In feature extraction phase two types of features, gradient features and density features were extracted and for the classification the gradient features were fed to a neural network and the density features were fed to another neural network and the two feed forward neural networks were combined using the combination schemes Max rule, Sum rule, Product rule and Borda count method. A dataset consisting of 825 samples written by 25 individuals was used for the experiment and they achieved 81.82% accuracy using the product rule combination scheme [25]. Varghese et al proposed a three stage feature extraction technique for recognizing handwritten character in Malayalam language.…”

Advancing Ensemble Learning Performance through data transformation and classifiers fusion in granular computing context

“…These rules have been popularly used in pattern recognition (Mangai et al, 2010;Kamble and Kokate, 2017), e.g. handwriting digits recognition (Shukla and Pandey, 2014), characters recognition (Chackoa and P.M.Dhanya, 2015) and affect recognition (Gunes and Piccardi, 2005).…”