On-line signature verification method utilizing feature extraction based on DWT

Abstract

In this paper, we propose a new signature verification method. Parameters of the on-line signature are decomposed into multilevel signals by utilizing the DWT (discrete wavelet transform). Personal features are emphasized by the DWT sub-band decomposition and extracted. The extracted features are verified using the adaptive algorithm. We also propose a method which is robust to fluctuation of the number of strokes by using DP (dynamic programming) matching. Through computer simulations, the effectiveness of this proposed method is confirmed.