Handwritten Digit Recognition using Adaptive Neuro-Fuzzy System and Ranked Features

This paper investigates Adaptive Neuro-Fuzzy Inference System (ANFIS) for recognition of handwritten digits. First, an efficient feature extraction module based on five feature extraction techniques has been performed. Second, an optimal feature selection method for feature ranking and feature reduction has been proposed. Third, a classification based on ANFIS has been done. The Experiments has been performed on standard handwritten digit dataset to evaluate the performance of the proposed system. Simulation result revels the proposed system has low testing and checking error with high recognition accuracy.