A Nonparametric Functional Method for Signature Recognition
Gery Geenens · Contributions to statistics · 2011
We propose to use nonparametric functional data analysis techniques within the framework of a signature recognition system. Regarding the signature as a random function from $$ \mathbb{R} {\rm(time \,domain)\,to}\, \mathbb{R}^2$$ (position (x,y) of the pen), we tackle the problem as a genuine nonparametric functional classification problem, in contrast to currently used biometrical approaches. A simulation study on a real data set shows good results.