Online Signature Verification Using a Single-template Strategy with Mean Templates and Local Stability-weighted Dynamic Time Warping

This study proposes a novel single-template strategy that uses mean templates and local stability-weighted dynamic time warping (LS-DTW) as a means of improving the speed and accuracy of online signature verification. Specifically, we adopt a recent time-series averaging method, Euclidean barycenter-based DTW barycenter averaging, to obtain effective mean templates while preserving intra-user variability among reference samples. Then, we estimate the local stability of the mean template set using multiple matching points that detect significant distorted trajectories in the warping paths of DTW. Subsequently, to boost discriminative power in the verification phase, we use the LS-DTW distances that incorporate the local stability sequence as the weights for the cost function of DTW warping between the set of mean templates and a test sample. Experimental results confirm the effectiveness of the proposed method using a common SVC2004 Task2 dataset.