私立东海大学资讯工程和科学研究所.PDFVIP

私立東海大學資訊工程與科學研究所 碩士論文 指導教授：周忠信 手寫數位簽名的生物特徵研究 The Study of Biometrics for Digital Handwriting 研究生：詹焯然 中華民國九十五年三月二十二日 摘要 本論文提出一個結合質心向量、速度、軌跡方向改變以及簽名大小-時間比 例值等生物資訊的手寫數位簽名相似度計算方法。利用生物特徵相似度，此方法 可以用來防止現行數位憑證認證中，數位憑證被竊取或被誤用的情況發生。與其 他生物資訊應用相比，手寫簽名生物資訊是一個行為上的生物特徵，因此，即使 複製簽名軌跡，但沒有即時簽名發生，也將會被分辨出來。而更重要的是，沒有 合法簽名者的同意，這些生物特徵也很難被獲得。在本研究中，當簽名的相似度 在80%以上時，我們即可認為這個擁有數位憑證的使用者就是合法的使用者。除 此之外，此系統同時也很容易的被實作或建置成網路上的應用系統。 關鍵字：認證、生物特徵、手寫簽名、相似度 i Abstract In this research, by combining the centroid, the speed, the stroke direction change and the size-time ratio of a handwriting trajectory as the biometrics of a person’s handwriting signature, an economic method to calculate the biometrics similarity between two handwriting signatures is proposed. The biometrics similarity can be used to prevent the situations such that either the digital certificates are stolen and misused or the users are commandeered illegally. To compare with other biometrics solutions, handwriting signature biometrics is a better candidate to be used in the Internet world since it is a behavior type biometrics and therefore duplicated signature without similar behavior is distinguishable and most importantly it can not be obtained without the person’s freewill. The experiments of our solution show that when digital certificate is verified and the returned biometrics similarity is higher than 80% it almost always can guarantee that the person is really own the certificate. The method proposed in this paper can be easily implemented and deployed for the Internet applications. Keywords: