Study On Face Recognition Based On Texture Features And Its Application On Android Platform

Compared with other biological recognition technologies,face recognition has some features like safe,reliable,unique and high user acceptance.Users can be recognized without contact with the device,therefore,it has a wide range of application scenarios.But face recognition easily affected by the factors such as illumination,posture,facial expression,especially the change of light conditions can make dramatic changes in the face image,and the difference between different light conditions is often greater than the difference between different faces.Therefore,study on face recognition under complex illumination has.the vital significance,and it is currently a hot topic in the field of face recognition.This paper mainly studies about complex illumination face recognition methods and its application on Android platform based on Local Binary Pattern(LBP).Face recognition includes image pretreatment,face detection,feature localization,feature extraction and recognition,etc.For the low rate and time-consuming of face recognition under complex illumination,this paper researches on video face detection algorithms under complex illumination and feature extraction algorithms base on Local Binary Pattern(LBP),and researches on apply face recognition to Android system to managing user’s permissions.The main research work includes the following three aspects:(1)Research on existing lighting preprocessing algorithms and use illumination invariant feature extraction algorithm based on Retinex theory for video image preprocessing.On this basis,AdaBoost face detection algorithm is improved.The extracted illumination invariant features are used to train AdaBoost classifier.Then the classifier is used to detect human faces.The experimental results show that this method can effectively improve the human face detection effect under complex illumination.(2)Based on illumination invariant features extracted in this paper,a new block weighted LBP facial feature extraction algorithm which fusions illumination invariant feature and original image is proposed.In this algorithm,the weight is determined by the light conditions and the importance of facial features.In order to verify the validity of this algorithm,a face detection and recognition system on the Windows platform is been created.The experimental results show that this algorithm has a higher recognition rate and faster recognition speed under complex illumination.(3)The algorithm realized in the Windows platform is transplanted to Android platform by using the JNI mechanism.For the privacy security issues of mobile devices,a new idea that combines face recognition with Android user’s access management is proposed.An Android application is designed to manage user’s permissions.It uses face recognition to determine whether the user is administrator or guest,then limit the guest’s permissions.