基于ARM的嵌入式人脸识别系统设计及实现
发布时间：2019-03-11 18:41
【摘要】：随着智能化信息处理技术的不断发展,人脸识别已成为身份认证领域的主要研究方向,并广泛应用于人们的日常生活。传统的人脸识别系统以电脑为平台,体积大,使用不灵活,而嵌入式ARM平台具有体积小、低功耗、高性能、低成本的特点,因此基于ARM平台开发人脸识别系统具有重要的应用价值。本文设计实现了一种稳定、准确的嵌入式人脸识别系统。首先,系统采用基于Crotex-A9架构的三星Exynos 4412开发板作为嵌入式ARM的开发平台,在此平台上完成了嵌入式Linux操作系统、Qt图像化界面和OpenCV计算机视觉库的移植。其次,深入分析并实现了基于Adaboost算法的人脸检测;实现并对比了基于Eigenfaces、Fisherfaces和LBPH三种算法的人脸识别;通过深入研究人脸图像预处理方法,从而进一步提高了识别性能。最后,本文设计出完整的嵌入式人脸识别系统,实现了基于USB摄像头的视频采集显示和人脸识别功能,并测试了系统四个功能模块,包括人脸库的建立、特征库的训练、导入配置文件和人脸识别。系统设计完成后,本文测试了系统的识别率和运行速度等性能指标。整个系统的测试结果表明本文设计的人脸识别系统具有较完备的功能、友好的界面、便捷的使用、高效的识别效果。
[Abstract]:With the development of intelligent information processing technology, face recognition has become the main research direction in the field of identity authentication, and widely used in people's daily life. The traditional face recognition system takes computer as the platform, the volume is large, the use is inflexible, and the embedded ARM platform has the characteristics of small size, low power consumption, high performance and low cost. Therefore, the development of face recognition system based on ARM platform has important application value. A stable and accurate embedded face recognition system is designed and implemented in this paper. Firstly, the system adopts Samsung Exynos 4412 development board based on Crotex-A9 architecture as the development platform of embedded ARM. On this platform, the embedded Linux operating system, Qt image interface and OpenCV computer visual library are transplanted. Secondly, in-depth analysis and implementation of face detection based on Adaboost algorithm; achieve and compare the three algorithms based on Eigenfaces,Fisherfaces and LBPH face recognition; through in-depth study of face image preprocessing methods, so as to further improve the recognition performance. Finally, this paper designs a complete embedded face recognition system, realizes the functions of video capture and display and face recognition based on USB camera, and tests the four functional modules of the system, including the establishment of human face database and the training of feature base. Import configuration files and face recognition. After the design of the system, the recognition rate and running speed of the system are tested. The test results of the whole system show that the face recognition system designed in this paper has more complete function, friendly interface, convenient use and high efficiency recognition effect.
【学位授予单位】：西安科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.41
本文编号：2438522
[Abstract]:With the development of intelligent information processing technology, face recognition has become the main research direction in the field of identity authentication, and widely used in people's daily life. The traditional face recognition system takes computer as the platform, the volume is large, the use is inflexible, and the embedded ARM platform has the characteristics of small size, low power consumption, high performance and low cost. Therefore, the development of face recognition system based on ARM platform has important application value. A stable and accurate embedded face recognition system is designed and implemented in this paper. Firstly, the system adopts Samsung Exynos 4412 development board based on Crotex-A9 architecture as the development platform of embedded ARM. On this platform, the embedded Linux operating system, Qt image interface and OpenCV computer visual library are transplanted. Secondly, in-depth analysis and implementation of face detection based on Adaboost algorithm; achieve and compare the three algorithms based on Eigenfaces,Fisherfaces and LBPH face recognition; through in-depth study of face image preprocessing methods, so as to further improve the recognition performance. Finally, this paper designs a complete embedded face recognition system, realizes the functions of video capture and display and face recognition based on USB camera, and tests the four functional modules of the system, including the establishment of human face database and the training of feature base. Import configuration files and face recognition. After the design of the system, the recognition rate and running speed of the system are tested. The test results of the whole system show that the face recognition system designed in this paper has more complete function, friendly interface, convenient use and high efficiency recognition effect.
【学位授予单位】：西安科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.41
【参考文献】
相关期刊论文 前6条
1 杨浩;张二喜;蒋卓芸;;基于距离测度的PCA人脸识别研究[J];陕西理工学院学报(自然科学版);2016年04期
2 韩艳斌;于威威;;基于LDA算法的人脸识别的研究[J];现代计算机(专业版);2015年18期
3 徐晓艳;;人脸识别技术综述[J];电子测试;2015年10期
4 田强;张权;朱桐;田青;夏卫军;戚鹿宁;;3D人脸识别技术及应用[J];警察技术;2014年05期
5 毛亮;李立琛;;浅谈人脸识别技术[J];中国公共安全;2014年10期
6 薛冰;郭晓松;蒲鹏程;;人脸识别技术综述[J];四川兵工学报;2010年07期
相关硕士学位论文 前1条
1 汪海波;基于深度学习的人脸识别方法研究[D];合肥工业大学;2014年
，本文编号：2438522
本文链接：https://www.wllwen.com/kejilunwen/ruanjiangongchenglunwen/2438522.html
