Signature verification using Particle Swarm Optimisation

Abstract

Signature is a typical method in authenticating identities. It is commonly used in many official transactions as a symbol of approval and agreement between all parties involved in the specified transactions. Although many individuals can possess the same name in their birth certificates, their signature still differs from one person to another. Therefore, this study examined the degree with which a signature is considered similar and belonged to the same person. The study focused on the online signature that is represented by (x, y) coordinates. We applied Particle Swarm Optimisation (PSO) to identify the baseline feature and proceeded to verify whether two signatures are similar to each other. The result of the baseline feature was then compared to the truth baseline and PSO was shown to have 90% accuracy in detecting whether it was a straight, ascending or descending baseline. We then tested the verification algorithm on the same person and then on different persons with the results showing an 80% accuracy.