| Project/Area Number |
12650382
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
情報通信工学
|
|---|
| Research Institution | Tottori University |
|---|
Principal Investigator |
NAKANISHI Isao Tottori University, Faculty of Education and Regional Sciences, Associate Professor, 教育地域科学部, 助教授 (80243377)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
ITOH Yoshio Tottori University, Faculty of Education of Engineering, Associate Professor, 工学部, 助教授 (70263481)
|
|---|
| Project Period (FY) |
2000 – 2002
|
| Project Status |
Completed (Fiscal Year 2002)
|
| Budget Amount *help |
¥1,900,000 (Direct Cost: ¥1,900,000)
Fiscal Year 2002: ¥400,000 (Direct Cost: ¥400,000)
Fiscal Year 2001: ¥500,000 (Direct Cost: ¥500,000)
Fiscal Year 2000: ¥1,000,000 (Direct Cost: ¥1,000,000)
|
|---|
| Keywords | on-line signature verification / wavelet transform / adaptive signal processing / stroke matching / dynamic programming method / ウェーブレット変換 / 適応フィルタ / 多次元処理 / 2次元処理 |
|---|
| Research Abstract |
The purpose of this research is to improve ability of on-line signature verification by introducing of the time-frequency analysis based on the wavelet transform. The results were as follows:
1. Daubechies' mother wavelet is suitable for the analysis.
2. Normalization of the data sample is necessary for independence of the input.
3. Stroke matching based on the dynamic programming (DP) method between the input data and reference one at each level is introduced.
4. When the difference of the number of the strokes within +__-2, the input is treated as a candidate of the genuine.
5. Reference data is produced by an average of five genuine data at each level.
6. Fluctuation of the amplitude at levels from 1 to 4 is very large, so that it is not suitable for verification.
7. Fluctuation of the amplitude at levels from 9 to 10 is quite small, so that is not suitable for verification, either.
8. For speed-up of convergence, the normalization of the step size in the adaptive algorithm is required.
9. A simple verification method based on the value of convergence, that is, the adaptive weight converges on 1 if the signature is of the genuine. Otherwise, it converges on 0, is achieved.
10. Total verification is achieved based on whether an average of the converged values of the upper 4 levels is larger than the threshold value or not.
As a result, we obtained 92% recognition rate when the threshold was 0.2.
|
