Off-fine Signature Recognition by Integrated Neural Networks

The purpose of this thesis is to propose an off-line automatic recognition system based on an Integrated Neural Networks (INN) which is composed of three modules, namely, Feature-distributed Network (FDN), Neural Network Classifier (NNC) and Decision-fusion Network (DFN). In the system, a specific classifier was created for each signature. A recognition training set was constructed for each classifier in order to solve the problems of off-line signature recognition and verification. Finally, the theoretical and practical possibilities were verified through the experiment conducted in the thesis.