The IOS Application Research Based On Face Recognition

Along with the rapid development of the Internet, the mobile communication and mobile terminal technology, the function of the smart phone is getting really stronger than the past years. Thus, the role it plays in human beings daily life is becoming more and more important. Nowadays, using mobile phones to log on the QQ or search the Internet are becoming customary. In recent years, some important functions, such as mobile payment and mobile bank, began to appear on the smart phones. With so much important information on the phone, the security of it is significant which attracts lots of attention.Using four numbers as the password and using a simple graphic pattern are the two popular ways of authentication. The former is usually for the iOS device and the latter for the Android device. However, both the two methods can be easily decoded by others. The new iPhone5s brings, a more credible identity authentication method, the fingerprint identification. Though the new method is quite safety, it has some shortage for its extra sensor and not open to developers.According to the above problems, this paper designs a facial recognition framework for iOS applications. In this framework, we use the camera which is common on iPhone to do identity authentication base on image processing techniques. Firstly, this dissertation introduces our topic’s research background and significance and the current research status around the world. Secondly, this dissertation presents the techniques and tools that are used in the system. We use Xcode to develop iOS applications with the Objective-C language and use OpenCV to do image procession. After that this dissertation analyses the three methods, Eigenfaces method, Fisherfaces method and LBPH method, which are available in OpenCV to do face recognition. Besides, we test all the methods on ORL and Yale face database and analyses the experimental result. Through the experiment, we found that Fisherfaces method is better than Eigenfaces and LBPH and with more than four pictures to be trained the recognition rate is acceptable. Then this dissertation introduces all the functions in the application and realizes them in five modules. Our application combines all the five modules to contain all the basic functions. Finally, this paper tests the application on iPhone5s. The application works well and realizes all the functions.