戴口罩多人脸识别系统研究

Research on multi-face recognition system with masks

针对新冠疫情及流感流行时期传统的人脸识别系统无法准确识别戴口罩人脸的问题,设计了一个口罩遮挡情况下的多人脸识别系统。该系统利用CLAHE算法对人脸图像进行预处理,通过Dlib库与Dnn模块定位人脸特征点,并对戴口罩人脸进行检测与识别,基于ResNet残差网络训练与质心追踪算法等技术提高检测效率。实验结果表明:本文所设计的戴口罩人脸识别系统成功率高,检测速度快,在多人脸识别应用中更能显示其优势。

In the epidemic period of the COVID-19 and influenza,traditional facial recognition systems can not accurately identify faces wearing masks,and therefore,a multi-face recognition system is designed to solve this problem.The face image is preprocessed by the CLAHE algorithm.By employing Dlib library and Dnn module,facial feature points were located,and the faces wearing masks were detected and recognized.The detection efficiency is improved based on ResNet residual network training and centroid tracking algorithm.Experiments show that the mask recognition system designed in this project has a high success rate and fast detection speed,which can show its advantages in multi-face recognition applications.