An automatic fuzzy neural network driven signature verification system

Abstract

An automatic fuzzy neural network driven signature verification system is developed in this paper. As Plamondon and Lorette (1989) have stated, the design of a signature verification system generally requires the solution of five types of problems: data acquisition, preprocessing, feature extraction, comparison process, and performance evaluation. However, unlike most existing automatic signature verification systems which employ traditional techniques (i.e. image processing techniques) to solve these problems, the proposed system is constructed on the basis of a novel fuzzy neural network named the pseudo outer-product based fuzzy neural network (POPFNN). The characteristics of the POPFNN, such as the learning ability, generalization ability, and high computational ability, make the signature verification system particularly powerful when verifying skilled forgeries. To test the efficacy of the proposed system, several kinds of POPFNNs are investigated in this paper. Their experimental results and comparisons are presented at the end of the paper for discussion.