應用影像辨識技術於NFC手機之研究

Abstract

隨著科技日益進步，智慧型手機功能越來越多樣化且應用範圍也越來越廣泛，而行動商務也隨之興起。把手機當作電子錢包、信用卡甚至透過智慧型手機來購物，已經是時代的趨勢。而如何有效且有鑑別性的進行身份驗證，是本研究致力解決的問題。 本研究以Android平台為基礎，利用NFC技術開發一套行動付款機制，並提出一個新穎的方法：以簽名的方式做身份的辨認，並利用影像處理技術辨識簽名來驗證使用者之身份，用改進以往手機做身份認證時，只能利用輸入PIN碼來做身份確認的不足。本研究使用Zernike moments技術來辨識簽名，此技術擁有抗旋轉不變性、抗位移不變性、抗大小不變性及能夠擷取出高鑑別度的特徵等特性，所以經實驗證明，Zernike moments技術確能運用在手寫簽章上的認證。

With the advance of information technology, the functions of smart phone are more diversified and its applications also are increased. The trend of using mobile phone as an electronic wallet, credit card, and even the shopping tools has arrived owing to the development of NFC (Near Field Communication). One of the most important modules of above applications is to verify the user’s authentication effectively and discriminatively. Current mobile payment mechanism only uses the PIN number to identify the user’s authentication. It is not friendly for that user need to remember lots of codes to keep the security of private profile. To solve this defect, we propose a novel approach to identify the user’s signature authentication. In this article, we develop a NFC mobile payment system which is based on Android operating system, and recognize the user’s signature with the Zernike moments technique. Owing to the merits of the Zernike moments technique, it can recognize the user’s signature with rotation invariance, transition invariance, and scaling invariance. Furthermore, it also can extract the high- discriminative features. From the result of our experiment, we find that the Zernike moments technical works well in the handwriting signature identification procedure.