Research Of Face Recognition System Under Complex Illumination Based On The Android Platform

This paper mainly studies the face recognition technology under complex lighting conditions,and the core technologies of face recognition are studied and analyzed in the platform of Android.System made a detailed research for face recognition system of preprocessing,feature extraction and classification recognition three key ring,and the performance of the system after each stage of processing will be improved.This design system mainly carries on the following research:First,This paper adopted Retinex algorithm in the pretreatment stage,analyses the intrinsic meaning of the Retinex theory and gets the conclusion of that the Retinex output image is essentially the relative reflection rate.As the relative reflection rate is insensitive to illumination,the Retinex theory can be used in face recognition under complicated illumination conditions.The histogram of the Retinex output images is calculated and close to normal distribution.So the Retinex output images mean and variance are normalized to get the intensity in display domain.The experiments indicate that this preprocessing algorithm can improve the robustness of face recognition against illumination,especially under uncontrolled coditions.Second,A variable light face recognition algorithm involved three points based on local binary pattern(LBP)and linear regression model is studied in the 4th chapter.It can obtain better performance,through processing the illumination at each stage.Then block LBP is applied to extract invariant feature which is robust to variant illumination.Finally,improved linear regression models used to classification which drops the first principal components.The proposed approach can reduce the effect of illumination,and improve robustness and the recognition accuracy of face recognition system.Third,This paper studies a method of human face detection based on OpenCV(Open Source Computer Vision)in Android system.The invoking and compliling of OpenCV functions with JNI(Java Native Interface)and Android NDK(Native Development Kit)is introduced in detail and shared library is got.Using JNI and OpenCV writting face recognition algorithm,and study it under the Android platform.