MODI Script Recognition System Using Zernike Moments and DCT Features

Modi Script or the old Marathi script was the script used to write Marathi language prominently before 1950's. Due to its intricate writing style and bent character properties, it has lost its meaning. In recent years, it has become crucial to examine old writings in order to glean meaningful information from them. It is difficult to recognize the documents since they are damaged and because the MODI script's overlapping characters make it tough. Different classifiers, including Decision Tree, SVM, KNN, and others, have been used in some MODI script recognition systems (MSRS) that have been proposed thus far. In order to achieve high accuracy, this research offers a Modi script recognition system that calculates Zernike moments and DCT coefficients as features. Each of the 27 classes in the custom dataset, which each represent a different character, was made. Different categorization methods, including Decision Tree, Random Forest, and KNN, were used. These were used to evaluate the system's performance. 97.77% high accuracy was achieved following feature fusion.