ANN-based handwritten character recognition

Abstract

Based on an artificial neural network, digital image processing, and features extraction theory, the authors analyzed a BP network's defect then presented improving solutions. In this paper, a new kind of handwritten character system has been constructed. Referring to the shortcoming of the traditional BP algorithm, a modified learning factor with adaptation is introduced, and a bizarre sample feature database is constructed for speeding up modified BP learning and classification. Experimental results show that the modified BP neural network algorithm (three layers forward, no feedback) can be used in handwritten character recognition, and satisfactory results have been obtained.