Multiple Local Weighting Scheme for Dynamic Time Warping in Online Signature Verification
Manabu Okawa · 2022 IEEE 4th Global Conference on Life Sciences and Technologies (LifeTech) · 2022
The world is witnessing a proliferation of technology, electronic applications, and products. Therefore, to stay away from any mishap, there is an unprecedented emphasis on cybersecurity. This paper proposes a novel online signature verification method that adopts a multiple local weighting scheme for dynamic time warping (DTW) in a single-template strategy to improve accuracy. In the enrollment phase, we created the mean templates of references using a time-series averaging method, i.e., the Euclidean barycenter-based DTW barycenter averaging. Further, we acquired multiple local weighting estimates for DTW between the mean templates and references for intra-user variability, followed by computing a global weighting estimate using gradient boosting for inter-user variability. Finally, we conducted the single-template strategy that applies the local and global weighting estimates to acquire a discriminative distance measure between a query sample and the mean template set. The experimental results using a public SVC2004 Task1 dataset showed the effectiveness of the proposed method.