Genetic algorithm based signature training

Abstract

Signature recognition has many practical applications in criminology, document authenticity, banking sector and organizational level record keeping. A trained template of sample input signatures can be used to recognize signatures by using a mapping technique. Therefore the signature template training process is very vital. In this paper we proposed a signature template training method based on Genetic Algorithm (GA). The training process mainly consists of two phases, signature preprocessing and GA based signature template training. Signature preprocessing stage includes signature acquisition using a scanner, signature background dependency adjustments, scale and translation invariance implementation and input signature image conversion to binary format.