Design And Implementation Of Android Platform Based On Face Recognition System

With the increase in computing and improve the processing performance of the smartphone storage space in the mobile-side implementation of face detection and recognition is no longer just an imagination. Currently, the phone automatically complete personal identity authentication has become a reality.This paper mainly focus on face recognition technology to achieve the study,, detailed study and analysis of the key technologies of Android system architecture and face recognition, face recognition based on theoretical knowledge and Android system proposed to achieve in Android platform Face recognition function design, and ultimately of the program, the main work done as follows:(1) by reading a lot of literature, complete the basic theoretical preparation work, summed up the development of face recognition system, face recognition and research status are introduced.(2) through the Android system in-depth study and research, the system architecture of the Android system development process and the Android platform in detail.(3) elaborated on the main process of face recognition, and face detection, feature extraction and recognition of the three main processes in the current mainstream method of a brief introduction. LDA on AdaBoost detection method and identification method used in this paper to study describes in detail the principle and the process steps of these two methods.(4) Based on the idea oflinking theory with practice, and based on the recognition of knowledge and research on the Android system, designed and implemented a set of face recognition system based on the Android platform. Make full use of the Android SDK and NDK development kit and resources, in the Windows system use Android Studio as a development environment, the Android platform a face recognition system.Experimental results show that, although the stability and accuracy of the system is relatively high, but because of the configuration of the device is relatively low, the real-time is not ideal, needs further research and exploration.