基于Face++的嵌入式实验室门禁人脸识别  

Face recognition of embedded laboratory entrance guard based on Face++

作　　者：李小华[1] LI Xiaohua(Yan'an Vocational and Technical College,Yan'an,shanxi 716000,China)

机构地区：[1]延安职业技术学院,陕西延安716000

出　　处：《自动化与仪器仪表》2021年第10期161-164,共4页Automation & Instrumentation

基　　金：陕西省技术创新引导计划“远距离人脸识别中低质图像增强关键理论研究”(No.2019JQ-907)。

摘　　要：为了提高实验室安全管理的工作效率,设计了一种基于Face++和嵌入式技术的人脸识别门禁系统。首先,对半自动触发模式的门禁系统总体架构进行设计,采用基于ARM9内核的S3C2440微处理器,并选用OV13850芯片的摄像头模块实现图像采集。其次,人脸检测与识别功能采用了开源的Face++云服务平台,并对人脸识别算法进行了改进,提出了一种动态调节阈值方案来提高识别的效率和准确率。通过FERET人脸数据集和AT&T人脸数据集对改进的Face++人脸识别算法进行了有效验证。实际测试结果表明,该实验室门禁人脸识别系统具有较高的准确和效率,平均识别率达95%以上。In order to improve the efficiency of laboratory safety management, a face recognition access control system based on Face++ and embedded technology is designed. At first, the overall architecture of the entrance guard system with semi-automatic trigger mode is designed, and S3 C2440 microprocessor based on ARM9 core is adopted, and the camera module of OV13850 chip is selected to realize image acquisition. Secondly, the face detection and recognition function adopts the open source Face++ cloud service platform, and improves the face recognition algorithm, and proposes a scheme of dynamically adjusting the threshold to improve the recognition efficiency and accuracy. The improved Face++ face recognition algorithm is effectively verified by FERET face dataset and AT&T face dataset. The actual test results show that the system can recognize faces accurately and efficiently, and the average recognition rate is over 96%.

关 键 词：实验室门禁 人脸识别 Face++ 动态阈值 ARM9 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]