On-Line Signature Handwritten Verification Algorithm Based on the Analysis of Frequency Range
Siliang Ma · Journal of Shenyang Jianzhu University · 2008
In order to improve the speed of online hand-written signature authentification,we propose an algorithm based on the analysis of frequency range,for the rough classification of online hand-written signature.Initially,we mapped the characteristic vectors of signatures to frequency range by the Fast Fourier Transform.Then we formed the new characteristic descriptions by extracting the part in low frequency.Finally,we achieved on-line hand-written signature verification with weighted verification algorithm.The algorithm improved the speed of online hand-written signature verification and the experimental result of ERR with the dataset of SCV2004 task 2 was 10%.The method,extracting frequency range characteristic using fast Fournier transformation,can be easily used to differentiate forge signatures,and it has great advantage in speed and efficiency.According to the stable degree of each kind of characteristics for each user,weighted authentification algorithm enhances the robustness of the system.