Implementation of an accurate Farsi Handwritten digit recognition system on FPGA

Abstract

An accurate Farsi Handwritten digit recognition system based on FPGA (Field Programmable Gate Arrays) implementation is proposed in this paper. For obtain high accuracy, we used a combined method for feature extraction. This method of feature extraction is appropriate for FPGA implementation as it can be implemented only with add and subtract operations. This greatly speeds up the process. This method is used to extract the features from normalized 40 × 40 pixel handwritten digit images. Applying this method to a Farsi digit recognition system is implemented using a two-layer MLP (Multi Layer Perceptron) artificial neural network with few neurons in the hidden layer. The system is simple, less complex and more accurate than the other similar systems.