Personal based authentication by face recognition Yung-Wei Kao, Hui-Zhen Gu, and Shyan-Ming Yuan Department of Computer Science and Engineering National Chiao Tung University, 1001 Ta Hsueh Rd., Hsinchu 300, Taiwan {ywkao, hcku}@cs.nctu.edu.tw, smyuan@cis.nctu.edu.tw Abstract Authentication is a significant issue in system control. Traditional account based authentication doesn’t guarantee the exact person of the account. It also suffers from the easily guessed problem. On the other hand, the camera is more and more popular. Personal based authentication via cameras becomes possible for mobile users. In the paper, we propose the PBAS, which conducts the authentication depends on both face image and password. We claim that the password information can be used to enhance face recognition rate, which is the most significant benchmark for identification system. Finally, we simulate the PBAS by PCA method based on both database we constructed and a subset of FERET. The result of experiment shows that our system performs much better than the PCA method without password integration. 1. Introduction Authentication is a significant issue in system control. Traditionally, the account based authentication is applied to recognize who is the current user. The users login to the system by their accounts and passwords. After the login, the consequent actions can be logged as belonging to a specific user. However, the account does not exactly represent the user himself. One user can own several accounts in a system, and one account can be used by several users. In many scenarios, it is significant to know who the user is, such as the critical business transactions. Many systems solve the problem by maintaining additional personal information while requesting the account. However, the account can still be used by those people knowing the password. Moreover, the account often composed by several characters or numbers, which is easily guessed. On the other hand, the camera is more and more popular. A lot of mobile devices and notebooks are equipped with built-in cameras. There are products (Lenovo Y510, IdeaPad Y510) even support face image recognition for authentication via cameras. The concept of face recognition for authentication has been implemented in products. However, most of them only depend on face images. We know that face image is public information; it is easy to be acquired. For example, we can login to the notebook owned by the president of the United States simply by showing his picture in front of the camera. Face recognition has been developed for a long time [1][2][3]. The most significant issue of face recognition is that the recognition rate is difficult to be 100%. The recognition rate of face recognition should be as high as possible. In our research, we propose the personal based authentication system (PBAS). PBAS conducts the authentication depends on both face image and password. PBAS has several advantages. First, the account factors of account based authentication are replaced by face images. Hence, the person who he is will be guaranteed. Second, only the face images are not enough to login, the passwords are still needed. Third, the accounts can be easily guessed, but the face images are not easily fabricated if you don’t know who the owner is. Finally, we will show that, the password is able to be utilized to enhance the face recognition rate. The paper is organized as followings: chapter 2 describes the system overview. Chapter 3 introduces the system design. Chapter 4 reviews the PCA method. Chapter 5 illustrates the evaluation of our simulation program. Finally, the conclusion is drawn. 2. System Overview Figure 1 represents the system overview of our research. The figure illustrates that a user is buying Fourth International Conference on Networked Computing and Advanced Information Management 978-0-7695-3322-3/08 $25.00 © 2008 IEEE DOI 10.1109/NCM.2008.167 581