筆跡情報と筆圧情報のHMMを用いたサイン認証
| Autor: | Kominami, Yoshifumi, Nishimura, Hiromitsu, Tomikawa, Takehiko |
|---|---|
| Jazyk: | japonština |
| Rok vydání: | 2006 |
| Předmět: | |
| Zdroj: | 神奈川工科大学研究報告.B,理工学編. 30:73-78 |
| ISSN: | 0916-1902 |
| Popis: | application/pdf In the field of user detection, detection methods using biometrical features have been reported in recent years. In this paper, we constructed an online sign detection system using a Hidden Markov Model (HMM). Constructing our user detection system is based on two types of HMM. One HMM is calculated from writing stroke features, another HMM is calculated from writing pressures. In this paper, our constructed HMM user detection system is compared with the sign detection system using simple pattern matching. A new sign database was gathered for the comparative experiments. The sign database consists of 800 sign by 4writers. In our experiment,the detection rate of our method performed 99.8%, which was about 3% better than the system using simple pattern matching. Moreover, HMM parameters for user detection were investigated from our experiment. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|