Feature extraction from on-line signature based on DWT and verification method using adaptive algorithm

摘要

We have proposed to decompose the time-varying pen-position parameters of on-line signature into multi level signals by using sub-band decompositions for the DWT (discrete wavelet transform). The high frequency components include individual features, which are emphasized through the sub-band decompositions. Also, we proposed a verification method based on the adaptive algorithm to the individual features. However, It is natural that there is fluctuation in the number of strokes in the on-line signature, so that some matching method of strokes is required to achieve higher accuracy. In this paper, we introduce a robust stroke matching by using the DP (dynamic programming) matching into our on-line signature verification method. Through computer simulations, the effectiveness of the proposed method is confirmed.