基于手机和人脸识别的身份识别系统

A NOVEL IDENTITY RECOGNITION SYSTEM BASED ON MOBILE PHONE AND FACE RECOGNITION

主要研究一种以具有摄像功能的手机端和人脸识别技术为基础的身份确认系统。利用局部二元模式(LBP)直方图特征的加权卡方距离来识别用户身份的方法,并根据手机用户的拍摄场景的多变性,添加对图片的光照预处理和几何校准过程,最后达到了比较好的实用效果。同时提出利用易辨识信息减少的比对样本,大大提高了系统运行性能。最后在ELSJTU图像数据集上测试的基础上,分析总结并提出了可以改进的意见。

In this paper we mainly researched on the design of an identity verification system based on mobile phone with camera and face recognition technology.We studied the approach of using weighted chi-square distance calculated from LBP histogram features to recognize the identities of users.According to the variability of shooting scenes the mobile phone users has,we appended a light pre-processing and geometrical calibration procedure,and these finally brought a satisfactory practical effect.Meanwhile we presented to use easily recognisable information to reduce the samples contrasted,which significantly improved the operation performance of the system.In the end,after testing based on ELSJTU image database,some improvements available were suggested based on the analyses and summarisation.