Design And Implementation Of Face Authentication System Based On Android

Nowadays,the rapid development of technology makes the function of mobile devices such as smart phones is no longer limited to some basic applications like text and call.Especially in recent years,the system's function and the hardware's performance of the mobile devices are both have a considerable degree of improvement,so the application on the Android platform is also a wide variety.In this case,more and more person attach importance to the security of mobile devices,and in the Android platform,these security issues relate to multiple contents,such as protecting the user's privacy and verifying the user's identity.When the mobile devices are lost or stolen,the user may not only lose a certain amount of property,but also leakage their private information.Once important information is obtained by criminals,it also can raise a series of serious consequence,such as defrauding user's relative of money by a text or call,stealing the user's property in the mobile applications and even using user's identity to crime.And now a lot of mobile applications judge the user on the basis of the password.The users have to set the password in a very complicated way in order to protect their private information,but this will inevitably lead to some unexpected circumstances,such as forgetting the password and losing the record.In view of the above problems,this paper proposes to achieve a face authentication system on the Android platform.In the first place,this system determines the contour of the complexion area according to the difference between the skin color and other colors,and clear the position and size of the face in the image.And then detect this image with the AdaBoost algorithm to further determine the exact location and the specific size of the face,which can eliminate the ill effects made by the non-complexion point.On the basis of the location of the feature points,the distribution and the change of the gray scale,this system propose the Haar feature which is more suitable for face detection get the classifier to improve the accuracy of identifying the face.In this process,this system obtains the image of faces with the camera and users can register in real time.This system also can accurately obtain the face area,so that it can extract the feature in the image which has little interference.The HOG feature can reflect all the information of the image in detail,so this system obtains the HOG feature vector in the first,and then reduce the characteristics of the vector with PCA.For the calculation of the gradient,this system add the use of the pixels on the diagonal,which can obtain the gradient of the center point more accurately.The face authentication system described in the paper can gather the face image in real time and identify the real user with high rate accuracy,so other persons can not obtain the user's information without permission.