Research On Fingerprint Anti-theft Identification Based On Android System

With the rapid development of communication technology,smart phones are an indispensable tool for people to go out,and the share of Android devices in China has exceeded 80%.Mobile phones have become the terminal of personal business,and people always store important personal information on their phones.If their phones are lost,it will not only cause social problems such as fraud and extortion,but also lead to economic crimes.According to the appeal issue,an effective and retrievable anti-theft method is currently needed.Therefore,this article proposes a fingerprint recognition method as an anti-theft method applied to Android phones.When the unlocking of the phone fingerprint fails,upload the fingerprint information attempting to unlock the phone to Raspberry Pi.Use image processing technology to filter out matching fingerprints in the self built fingerprint database,and send the bound identity information to the user’s bound email through Raspberry Pi.This fundamentally solves the problem of phone loss and improves the success rate of retrieving the phone.The main work and achievements are as follows:1.Implement image acquisition based on Raspberry Pi and ZW800 capacitive fingerprint module,and establish a fingerprint database within Raspberry Pi.The system mainly preprocesses,refines,and binarizes the collected fingerprints,and then extracts feature point coordinates and directions based on the MCC matching algorithm,which is stored in an array form in the database.2.Construct a minimum recognition area testing system for the ZW800 capacitive fingerprint module.By randomly obtaining cropped fingerprint images of different sizes and simulating the fingerprint information randomly collected by the fingerprint module on a mobile phone,the minimum area that the module can perform fingerprint recognition is tested.The experiment shows that when the collected fingerprint image area is 160*160 pixels,the minimum area that can be recognized is 40*40 pixels.3.In order to reduce the time spent traversing the fingerprint database,a large number of fingerprint images(divided into five types of labels)are used to train the AlexNet network,obtain a neural network model for fingerprint classification,and build it in a fingerprint recognition system.By using classification to establish various types of fingerprint databases,the total fingerprint database data is divided into five parts.The experiment shows that the recognition accuracy of the model after 20 iterations reaches 95.20%,which can reliably classify fingerprints.4.Write Android application software to achieve user registration,login,binding information,and uploading fingerprint functions.Use Raspberry Pi as the server end and Android phone as the user end,and achieve data transmission with Raspberry Pi’s fingerprint recognition system through network programming.The main content of the transmission is to attempt to unlock the fingerprint information of the phone.This information is transmitted in the form of a byte stream to the Raspberry Pi,decoded to obtain binary data,converted into matrix form,extracted feature point information through image processing technology,and screened in the fingerprint database.The experimental results show that the accuracy rate of screening and investigation reaches 96.89%,and the recognition rate of app fingerprint authentication is 0.