Spectrum Analysis Based onWindows with Variable Widths for Online Signature Verification

Abstract

In this paper, an online signature verification scheme based on spectrum analysis and Mahalanobis decision is proposed. We firstly divided signatures to a number of frames with variable widths according to the characteristics of the time sequences, and then employed the fast Fourier transformation (FFT) to extract the spectrum of signatures. The distance between the Fourier coefficient within the corresponding frames is computed, and the Mahalanobis decision making is employed. Experimentation demonstrates that spectrum analysis based on windows with variable widths is effective for online signature signals