CASIA OpenIR  > 毕业生  > 硕士学位论文
Alternative TitleEmbedded Face Recognition System Research and Development
Thesis Advisor李子青
Degree Grantor中国科学院研究生院
Place of Conferral中国科学院自动化研究所
Degree Discipline软件工程
Keyword嵌入式linux 人脸识别 达芬奇技术 Tms320dm6446 编码引擎 Embedded Linux Face Recognition Davinci Tms320dm6446 Codec Engine
Abstract随着社会科技发展及信息化速度不断加快,国家公共安全,信息安全等关系国计民生的领域亟需高可靠性,高安全性的全新身份识别技术。相对刷卡,口令和签名等易被复制、破解的传统身份验证方式相比,人脸具有难以被伪造,不会遗失,非接触性等特点。人脸识别在安全验证系统,通关口岸,关键区域门禁等方面有着巨大的应用前景。传统的人脸识别系统都是基于PC平台应用,但在实际应用中要求占据更小的空间,具有更高的便携性和易操作性,嵌入式技术的飞速发展促进了这一应用的实现。本文深入研究了嵌入式Linux系统软件,将人脸识别算法移植到TI双核异构的TMS320DM6446处理器上,最后,实现整个应用软件。论文主要的贡献如下: 1)深入研究了TI的双核异构软件基础平台Codec Engine的原理,探讨了片上双核TMS320DM6446中ARM和DSP数据交互方式。研究了DSP端XDM算法实现方法,并移植近红外人脸识别算法,运动检测算法和JPEG编码算法; 2)研究了嵌入式Linux操作系统;搭建了开发嵌入式系统的交叉编译环境,包括TFTP,nfs等网络工具;研究了bootloader,uImage等系统软件;移植了vsftp和http网络服务器;设计和制作了I2C键盘驱动和基于串口的遥控驱动; 3)深入研究了Davinci技术和OSD显示模块工作原理,使用DirectFB+LiTE实现应用软件的GUI,同时学习了V4L2采集模块,实现系统近红外和可见光的双路视频采集模块,使用gettext实现软件的国际化。最后,对应用软件进行优化。 该系统成功应用于2010年上海世博会第三安保责任区,得到广泛好评。通过项目实施,该系统表现出了较高的易用性,为后续进一步开发奠定了坚实基础。
Other AbstractPublic security, information security and people security demand new person identification with high reliability. Compared to traditional identity authentication technologies, such as RFID card, password and signature which easily be copied, decode, face biometric is difficult to forge, can not be lost, and is nonintrusive and easy to us. Face recognition in verification system, immigration clearance at customs, access control in important area and other aspects has huge application prospect. Recently, face recognition system is usually implemented on PC platform. Embedded system is a good solution, and usually in different kinds of environments. In this thesis, research is conducted to port face recognition algorithm to TI dual-core heterogeneous processor TMS320DM6446 and to implement the application software running on an embedded linux system. The main contributions as follows: 1) A deep research on the principle of Codec Engine of the TI dual-core heterogeneous software foundation, and probes data interactive mode between the ARM and DSP of the TMS320DM6446 on-chip system. The study of XDM algorithm realization method on DSP, and transplant near infrared face recognition, motion detection and JPEG encoded algorithms; 2) Learn the Embedded Linux operating system. Build the development of the embedded system environment, including cross-compilation, network tools, such as TFTP, NFS, uImage Boot-loader and other system software. To transplant VSFTP and HTTP web server to embedded linux platform. Design and make the I2C keyboard and the remote control driver based on serial driver; 3) A deep research on DaVinci technology and working principle of the OSD display module. Using DirectFB + LiTE realize the GUI of application software, and study V4L2 acquisition module, realize double road video acquisition module of near infrared and visible, using gettext realize internationalization. Finally, the application software is optimized. This system successfully applies to the third security area 2010 Shanghai world expo, and gets widely praise. Through the project implementation, this system shows a higher ease-of-use, last but not the least, it laid a solid foundation for the subsequent development of Omap3530.
Other Identifier200828009029042
Document Type学位论文
Recommended Citation
GB/T 7714
高凡. 嵌入式人脸识别系统开发研究[D]. 中国科学院自动化研究所. 中国科学院研究生院,2011.
Files in This Item:
File Name/Size DocType Version Access License
CASIA_20082800902904(2691KB) 暂不开放CC BY-NC-SAApplication Full Text
Related Services
Recommend this item
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[高凡]'s Articles
Baidu academic
Similar articles in Baidu academic
[高凡]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[高凡]'s Articles
Terms of Use
No data!
Social Bookmark/Share
All comments (0)
No comment.

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.