Complex Mathematical Expressions Recognition using Support Vector Machine as a Classifier

Abstract

Mathematics is almost unavoidable in daily life as well as in mathematical model with analysis in recent technology. It is today’s need to implement math recognition system to participate in digital world. Each and every element in the equations has been segmented using morphological operations and subsequently the features entropy, mean, variance, standard deviation, skewness, kurtosis, correlation, contrast  are extracted and used appropriate classifier- support vector machine to improve the accuracy .  The more complex handwritten equations have been considered for the recognition. The implemented algorithm can be used in offline recognition of equations, digits and symbols on postal and bank documents, university answer sheets, handwritten notes of mathematics as well as in blind math applications. The efficiency of the system is measured using confusion matrix and ROC (Receiver Operating Characteristics) under AUC (Area under Curve). Basically accuracy is depends on extracted features and classifier used.