THE DEMPSTER-SHAFER THEORY COMBINED WITH NEURAL NETWORK IN HANDWRITTEN CHARACTER RECOGNITION

A novel character recognition method, called the Dempster-Shafer theory combined with Neural Network in Handwritten Character Recognition (DSNNHCR), is proposed in this paper. The DSNNHCR integrates a Recurrent Neural Network (RNN) and the Dempster-Shafer theory (D-S) to recognize handwritten characters. It first employs an RNN to effectively extract oriented features of a handwritten character. Subsequently, the method creates 3 feature variables using extracted oriented features. Finally, 3 feature variables are applied to the Dempster-Shafer theory which can powerfully estimate the similarity ratings between a recognized character and sampling characters in the character database. Experimental results demonstrate that the DSNNHCR system achieves a satisfying recognition performance.