Wavelet Thinning Algorithm Based SimilarityEvaluation for Offline Signature Verification

摘要

Structure distortion evaluation is able to allow us directly measurernsimilarity between signature patterns without classification using feature vectorsrnwhich usually suffers from limited training samples. In this paper, we incorporaternmerits of both global and local alignment algorithms to define structure distortionrnusing signature skeletons identified by a robust wavelet thinning technique. Arnweak affine model is employed to globally register two signature skeletons andrnstructure distortion between two signature patterns are determined by applying anrnelastic local alignment algorithm. Similarity measurement is evaluated in the formrnof Euclidean distance of all found corresponding feature points. Experimentalrnresults showed that the proposed similarity measurement was able to providernsufficient discriminatory information in terms of equal error rate being 18.6%rnwith four training samples.