Signature Verification Based on a Fuzzy Genetic Algorithm
J.N.K. Liu, George S. K. Fung · 2020
The signature of an object can be regarded as some feature/characteristic specific to it, which shall be differentiated from that of other objects for verification purposes, including recognition, identification and authentication. This chapter analyses the important signature features and verify signatures with greater certainty. The use of fuzzy genetic algorithm (FGA) overcomes the traditional problems in feature classification and selection, providing fuzzy templates for the identification of the smallest subset of features. FGA can be applied to select the key features and the class-dependent discretization method to discrete the selected features. Many signatures are subject to two types of signature variability: intraclass variability and interclass variability. A facial signature is a set of facial features extracted from 400 different grayscale images of 40 different subjects. There are two major advantages of fuzzy genetic algorithm. First, the degree of importance for each feature within the key signature set could be evaluated. Second, the performance of the classifier could be improved.