Research On Palmprint Preprocessing In Complex Background And Its System Implementation For Mobile Phones

In recent years,with the rapid development of the mobile Internet,it has put higher requirements on the authentication security of the mobile terminal that online transactions,mobile payments applications or the like have appeared.At present,there are security risks in the case that the phone is lost or the password is forgotten because the authentication technology of the mobile terminal mostly uses passwords.Biometrics are considered to be effective means to solve the problem of identity authentication.Among them,palmprint has the characteristics of rich and stable textures and simple acquisition mode.Therefore,the biometric authentication technology of mobile terminal based on the palmprint has become the current research hotspot.Due to problems such as rotation,scaling,translation,changing of illumination,and complex background in the palm image captured by the mobile terminal,effectively segmenting palm images and locating the ROI of palmprint(Region of Interests)in complex background become the key steps of the palmprint authentication technology of the mobile terminal.The foreground segmentation of palm images and the locating of palmprint ROI in complex backgrounds are studied in this paper.The main contributions are as follows.(1)Aiming at the problem of palm image segmentation under complex background,an algorithm of palm image segmentation based on an improved active shape model is proposed.A priori knowledge of the palm shape is obtained from the palm image of the training set.In order to reduce the interference caused by non-shape factors such as palm stretching and bending or the like,different vectors of palm shape are uniformly aligned and registered using Procrustes analysis.The principal component analysis method is used to analyze the palm shape vector and establish a palm shape model.In order to reduce the interference of illumination,an improved local feature model is established for the feature points which form palm shape.The foreground of palm image is segmented by using the initialized palm shape model and local feature point's model.Contour extraction algorithm and convex hull theory are used to locate the key points and extract palmprint ROI.(2)A contactless palmprint authentication system on the mobile terminal was designed and implemented.Through the analysis of the requirements of the palmprint authentication system on the mobile terminal,the system architecture of the palmprint authentication system is designed.The system mainly consists of three modules:palm image acquisition module,registration module and authentication module,which realizes functions of palmprint registration and palmprint authentication.The palmprint authentication system under the Android platform is implemented by Android Studio 2.3,and the functional test of the system is carried out.The Palmprint authentication system of mobile terminal is of great value in solving the effective identity authentication in the process of mobile payment.