Research And Implementation Of Face Recognition Based On Android Platform

With the development of mobile Internet, smart phones have become indispensable to human daily learn and life and also has gone beyond its original design instinct. But numerous phone information security problems, such as mobile payment platform, the user’s private information administration, have also become eager to solve. It is known that biometric technology is a technique using human physiological or behavioral characteristics to authenticate the action and biometrics have strong stability and significant individual differences, so it is ideal for identity verification feature.Face recognition includes image preprocessing, face detection, feature extraction, recognition and classification. This paper researched the key issues of the face recognition, and achieved the image-based face recognition system on the Android platform. The main research contents are as follows:(1) The image preprocessing was studied, common and classic pretreatment methods, such as gray transform, histogram equalization, image smoothing, image sharpening, image binarization, are mainly analyzed and studied, and these preprocessing algorithm was given the process and effects processing.(2) The face detection method was analyzed, especially the mainstream Adaboost learning algorithm of face detection, AdaBoost algorithm based on Haar features was mainly researched. According to the existing literature, an improved Haar the Adaboost algorithm was analyze.(3) Feature extraction was an important step of face Recognition. The methods of facial feature extraction were analyzed and summarized, especially discrete cosine transform of frequency domain transform. The discrete cosine transform has many advantages, for example, error rate is small, the ability to focusing on energy is good, compression capability and computational complexity are better. So discrete cosine transform was suitable for the development of the Android platform. As references cited, herein facial feature extraction algorithm based on a combination of DCT and FLDA was cited. Finally, the feature extraction algorithm based on the minimum distance classifier was proven on the ORL face database. The experimental results showed the effectiveness of the algorithm.(4) The Android programming platform was analyzed, including the Android development, Android NDK, the construct of Android development platform. And the compiler libraries were analyzed, finally face recognition system was achieved on the Android platform system. The results showed that the system has good stability and effectiveness.