Browsing by Author "Wu, QZ"

Now showing 1 - 3 of 3

On-line signature verification based on logarithmic spectrum
(1998-12-01)
Wu, QZ
;
Lee, SY
;
Jou, IC
;
資訊工程學系
;
Department of Computer Science
In this paper, an on-line signature verification scheme based on similarity measurement of logarithmic spectrum is proposed. The principal components of the logarithmic spectrum of each signature are extracted. We then compute the similarity of logarithmic spectrum between input signature and the reference template. By comparing the similarity of logarithmic spectrum with the verification threshold, we can determine the authenticity of the input signature. Based on the experimentation, the rates of false rejection errors and false acceptance errors are as low as 1.4 and 2.8%, respectively. This demonstrates the effectiveness of the proposed scheme. (C) 1998 Published by Elsevier Science Ltd on behalf of the Pattern Recognition Society. All rights reserved.
On-line signature verification based on split-and-merge matching mechanism
(1997-07-01)
Wu, QZ
;
Lee, SY
;
Jou, IC
;
資訊工程學系
;
Department of Computer Science
In this paper, an on-line signature verification scheme based on split-and-merge matching mechanism is proposed. Each word in the signature is specified with static and dynamic features: a sequence of (x, y) coordinates and a sequence of (x, y) velocities. We employ the split-and-merge matching mechanism for each input sequence of coordinates or velocities with the corresponding sequence of the reference template. Alignment and refinement are exploited in the matching to justify the data skew problem. The accumulated coordinate and velocity distances are compared with corresponding verification thresholds to determine the genuineness of the input signature, Algorithms to compute reference templates and verification thresholds are also proposed. The performed simulation results support the effectiveness of this signature verification scheme.
On-line signature verification using LPC cepstrum and neural networks
(1997-02-01)
Wu, QZ
;
Jou, IC
;
Lee, SY
;
資訊工程學系
;
Department of Computer Science
In this paper, an on-line signature verification scheme based on Linear Prediction Coding (LPC) cepstrum and neural networks is proposed. Cepstral coefficients derived from linear predictor coefficients of the writing trajectories are calculated as the features of the signatures. These coefficients are used as inputs to the neural networks. A number of single-output multilayer perceptrons (MLP's), as many as the number of words in the signature, are equipped for each registered person to verify the input signature. If the summation of output values of all MLP's is larger than verification threshold, the input signature is regarded as a genuine signature; otherwise, the input signature is a forgery. Simulations show that this scheme can detect the genuineness of the input signatures from our test database with an error rate as low as 4%.