基于自适应LBP人脸识别的身份验证

提出了一种自适应LBP人脸识别算法用于进行身份认证。在身份特征录入阶段，首先采用Harr人脸级联分类器对人脸样本库进行人脸区域检测，并使用PCA方法对人脸区域进行降维处理；然后通过LBP二值模式的人脸识别算法提取人脸样本的特征值；最后通过LBP人脸训练生成人脸数据特征库。通过多场景人脸图像库和阈值队列，通过多阈值全组人脸匹配，建立人脸阈值特征库。在身份验证阶段，将登录用户人脸与人脸阈值特征库做粗粒度人脸LBP直方图匹配，确定当前最优的LBP阈值；然后将登录用户人脸与人脸数据特征库做LBP直方图匹配，通过匹配结果确定登录用户的系统权限。实验结果表明，在图像和视频模式下，基于自适应LBP人脸识别算法的身份验证具有很高的鲁棒性。

Identity authentication based on adaptive LBP face recognition

An identity authentication method based on adaptive LBP face recognition is proposed.In the identity feature recording phase,Harr face cascade classifier is first employed on face sample library to detect the face area,then PCA method is adopted to reduce dimensionality for face area,an adaptive LBP face recognition algorithm is adopted to extract the feature values of face samples,and at last these feature values are trained to construct the face feature library.The face threshold feature library is constructed by multi scene face image library,threshold queue,and multi threshold matching of faces in the whole group. In the identity authentication phase,the LBP matching method is used between the face of the certificating user and face threshold feature in the face threshold feature library so as to determine the best LBP threshold.Using this threshold,the face of the certificating user is matched with the face feature in the face feature library in order to confirm the system permission for the certificating user.The experiments show that the identity authentication method based on adaptive LBP face recognition is of higher recognition rate and robustness.