基于ARM11的嵌入式人脸识别系统的设计和实现
|
摘要
作为生物体特征识别技术,人脸识别技术具有使用方便、难以仿冒、识别精度高、速度快等特点。而传统的人脸识别系统都是在以PC为基础的平台上进行开发的,这样的系统在使用的场合和成本上都有很大的缺陷。而随着嵌入式技术的发展,现在短小精悍、方便实用、成本低廉的嵌入式系统平台的已经具有很强的数据处理能力,嵌入式人脸识别已成为目前研究和应用的趋势。
本文设计实现了一种准确性、稳定性较高的嵌入式人脸识别系统,为此主要做了以下几方面的工作：首先搭建了嵌入式系统平台。为了弥补了ARM9运算能力不强,ARM+DSP组合价格偏高的不足,本文选用以ARM11为核心的tiny6410开发板和普通USB摄像头作为系统的硬件平台,并在平台上移植了Linux操作系统、Qt图像化界面库和Opencv计算机视觉库。然后设计实现了基于此嵌入式平台的人脸识别系统。对系统图像采集及格式转换模块,人脸检测模块,人脸图像预处理模块和人脸识别模块所采用的方法和实现过程分别进行了详细介绍,并对系统性能做了测试。测试结果表明系统达到了预期的设计目标。最后针对人脸识别系统中反照片欺诈的问题,论文提出了一种基于眨眼检测的反照片欺诈方法和一种基于视线变化的反照片欺诈方法,并对所提方法进行了测试,初步实现了反照片欺诈的目标。
As a biological feature identification technology,face recognition technology has the advantages of convenient use, difficult to counterfeit,high precision and speed etc.. But the traditional face recognition systems are based on PC,this system has a lot of defects in the use of the occasion and cost. With the development of embedded technology.Now small, convenient and practical, low cost embedded system platform has strong data processing ability.Embedded face recognition will become the mainstream of future research.
The purpose of this paper is to design a face recognition with high accuracy and stability.For this,article has done the following work. The first is to build a system platform, In order to make up for the less operational capability of ARM9, and lack of high prices ARM+DSP system. This design uses a development board tiny6410based on ARM11and the USB camera as the hardware platform of the system. Then i have transplanted Linux operating system,Qt graphic interface and Opencv computer vision library. And then i introduce the implementation process of the embedded face recognition system.The principle,method and the implementation procedure of mage capture and format conversion module,face detection module,face image preprocessing module and face recognition module is introduced.And design of the whole system is completed.The system was tested,and the test shows that the expected design is achieved.At last for the photo deception problems.The anti-photo detection method which is based on blinking eye detection and the changing of eye vision are raised.And experiment is done and the expected result is obtained.The result shows that the system can distinguish the photo and the real person.
本文设计实现了一种准确性、稳定性较高的嵌入式人脸识别系统,为此主要做了以下几方面的工作：首先搭建了嵌入式系统平台。为了弥补了ARM9运算能力不强,ARM+DSP组合价格偏高的不足,本文选用以ARM11为核心的tiny6410开发板和普通USB摄像头作为系统的硬件平台,并在平台上移植了Linux操作系统、Qt图像化界面库和Opencv计算机视觉库。然后设计实现了基于此嵌入式平台的人脸识别系统。对系统图像采集及格式转换模块,人脸检测模块,人脸图像预处理模块和人脸识别模块所采用的方法和实现过程分别进行了详细介绍,并对系统性能做了测试。测试结果表明系统达到了预期的设计目标。最后针对人脸识别系统中反照片欺诈的问题,论文提出了一种基于眨眼检测的反照片欺诈方法和一种基于视线变化的反照片欺诈方法,并对所提方法进行了测试,初步实现了反照片欺诈的目标。
As a biological feature identification technology,face recognition technology has the advantages of convenient use, difficult to counterfeit,high precision and speed etc.. But the traditional face recognition systems are based on PC,this system has a lot of defects in the use of the occasion and cost. With the development of embedded technology.Now small, convenient and practical, low cost embedded system platform has strong data processing ability.Embedded face recognition will become the mainstream of future research.
The purpose of this paper is to design a face recognition with high accuracy and stability.For this,article has done the following work. The first is to build a system platform, In order to make up for the less operational capability of ARM9, and lack of high prices ARM+DSP system. This design uses a development board tiny6410based on ARM11and the USB camera as the hardware platform of the system. Then i have transplanted Linux operating system,Qt graphic interface and Opencv computer vision library. And then i introduce the implementation process of the embedded face recognition system.The principle,method and the implementation procedure of mage capture and format conversion module,face detection module,face image preprocessing module and face recognition module is introduced.And design of the whole system is completed.The system was tested,and the test shows that the expected design is achieved.At last for the photo deception problems.The anti-photo detection method which is based on blinking eye detection and the changing of eye vision are raised.And experiment is done and the expected result is obtained.The result shows that the system can distinguish the photo and the real person.
引文
[1]周杰,卢春雨,张长水等.人脸自动识别方法综述[J].电子学报,2000,28(4)：102～106
[2]Chellappa R, Wilson C L,Sirohey S. Human and machine recognize of faces:a survey.Proc.IEEE,1995,83(5):705-740
[3]Pentland T M. A eigenfaces for recognition [J]. Journal of Cognitive Neuroscience,1991,3(1): 71-86
[4]刘俊宁.基于LPP算法的人脸识别技术研究[D].[硕士毕业论文].镇江：江苏大学,2010
[5]尹克重.基于ICA的人脸识别算法研究[D].[硕士毕业论文].重庆：重庆大学,2006
[6]王海珍.基于LDA的人脸识别技术研究[D].[硕士毕业论文].西安：西安电子科技大学,2010
[7]何立民.嵌入式系统的定义与发展历史[J].单片机与嵌入式系统用.2004,1：6～8
[8]周激流,张哗.人脸识别理论研究进展；计算机辅助设计与图形学学报[J].第11卷,第2期,1999.
[9]孙朋,刘党辉,肖艳青。基于嵌入式系统的人脸识别技术研究进展[C].第七届全国信息获取与处理学术会议论文集
[10]姜钰,基于ARM9的人脸识别系统硬件平台设计[D].[硕士学位论文].南京：南京航空航天大学,2010
[11]刘立.基于双核架构的嵌入式人脸识别系统,[D].[硕士学位论文].上海：上海交通大学,2012
[12]王元伟,刘国秀.基于V4L的视频采集系统的设计[J],铁路计算机应,2010,7(19)：30-34
[13]Sumsung Electronics.S3C6410X RSIC Microprocessor User's Manual [K]. Sumung Electronics,2009
[14]刘文峰,李程远,李善平.嵌入式Linux操作系统的研究[J].浙江大学学报(工学版),2004,4：60～65
[15]刘磊,张凤荔,秦志光.基于U-boot构建嵌入式Linux的Bootloader[J].计算机应用研究,2007,12：238～240
[16]刘文,徐磊,盛文婷.基于ARM平台的Linux内核分析与移植研究[J].现代计算机(专业版),2011,7：72～75
[17]韦斯,丁志刚,张伟宏.LINUX下UBI子系统的研究与应用[J].2010,27(10)：68
[18]郭晖,陈光.基于OpenCV的视频图像处理应用研究[J].微型机与应用,2010,21(29)：14～17
[19]徐义鑫,韦青,尚进,程德芳.Qt在嵌入式系统中的移植研究与实现[J].电子技术,2008,9(45)：49～51
[20]徐家,陈奇.基于V4L2的视频设备驱动开发[J].计算机工程与设计,201016(31)：3569-3580
[21]Ming-Hsuan Yang, David J. Kriegman, Narendra Ahuja, "Detecting Faces in Images:A Survey"[C].IEEE Transaction on Pattern Analysis and Machine Intelligence, January 2002,1(24):34-58
[22]Viola P, Jones M. Rapid object detection using a Boosted cascade of simple features [C]. Proc of IEEE Conference on Computer Vision and Pattern Recognition. Kauai, Hawaii, 2001:511-518
[23]任艳斐.直方图均衡化在图像处理中的应用[J].科技信息.2007.4：37～38
[24]李建中,雷立礼,黎灏.基于几何特征的动态人脸识别[J].科学技术与工程.2010.28：6894～6899
[25]谢毓湘,王卫威,栾悉道.基于肤色与模板匹配的人脸识别[J].计算机工程与科学.2008,30(6)：54～56
[26]余龙华,王宏,钟洪声.基于隐马尔科夫模型的人脸识别[J].计算机技术与发展.2012,22(2)：25～29
[27]刘晓东.基于弹性图匹配法的人脸识别算法研究[D].[硕士学位论文].上海：上海交通大学,2009
[28]Pentland A,Choudhury T. Face recognition for smart environments [C]. IEEE Computer, 2000,50-55
[29]孟海东,张玉英,宋飞燕.一种基于加权欧氏距离聚类方法的研究[J].计算机应用,2006,s2：152-153
[30]T.Choudhu,B.Clarkson,T.Jebara.Multimodal person recognition using unconstrained audio and video. AVBPA'99,1999,176-181
[31]R.Brunelli,D.Falavigna.Personal Identification Using MultiPle Cues.IEEE Trans.Pattern Analysis and Machine Intelligence,1995,17(10):955-966
[32]J.Li,Y.Wang,T.Tan,and A.K.Jain.Live face detection based on the analysis of fourier spectra,SPIE.2004:296-303
[33]Klaus Kollreider,Hartwig Fronthaler,Mayeel Isaae Faraj.Real-time Face Deteotion and Motion Application in "Liveness" Assessment.IEEE Transactions on Information Forensics and Security,2007,2(3):548-558
[34]刘瑞安,靳世久,张希坤等.眨眼检测与眼睛跟踪[J].计算机应用.2006,26(12)：2835～2837
[35]高军,李学伟,张建.基于模板匹配的图像配准算法[J].西安交通大学学报,2007,41(3)：307～311
[36]朱桂英,张瑞林.基于Hough变换的圆检测方法[J].计算机工程与设计,2008,29(6)：1462～1465
[37]朱俊,任明武,杨章静.基于角点检测的快速匹配算法[J].南京理工大学学报,2001,35(6)755～759
[2]Chellappa R, Wilson C L,Sirohey S. Human and machine recognize of faces:a survey.Proc.IEEE,1995,83(5):705-740
[3]Pentland T M. A eigenfaces for recognition [J]. Journal of Cognitive Neuroscience,1991,3(1): 71-86
[4]刘俊宁.基于LPP算法的人脸识别技术研究[D].[硕士毕业论文].镇江：江苏大学,2010
[5]尹克重.基于ICA的人脸识别算法研究[D].[硕士毕业论文].重庆：重庆大学,2006
[6]王海珍.基于LDA的人脸识别技术研究[D].[硕士毕业论文].西安：西安电子科技大学,2010
[7]何立民.嵌入式系统的定义与发展历史[J].单片机与嵌入式系统用.2004,1：6～8
[8]周激流,张哗.人脸识别理论研究进展；计算机辅助设计与图形学学报[J].第11卷,第2期,1999.
[9]孙朋,刘党辉,肖艳青。基于嵌入式系统的人脸识别技术研究进展[C].第七届全国信息获取与处理学术会议论文集
[10]姜钰,基于ARM9的人脸识别系统硬件平台设计[D].[硕士学位论文].南京：南京航空航天大学,2010
[11]刘立.基于双核架构的嵌入式人脸识别系统,[D].[硕士学位论文].上海：上海交通大学,2012
[12]王元伟,刘国秀.基于V4L的视频采集系统的设计[J],铁路计算机应,2010,7(19)：30-34
[13]Sumsung Electronics.S3C6410X RSIC Microprocessor User's Manual [K]. Sumung Electronics,2009
[14]刘文峰,李程远,李善平.嵌入式Linux操作系统的研究[J].浙江大学学报(工学版),2004,4：60～65
[15]刘磊,张凤荔,秦志光.基于U-boot构建嵌入式Linux的Bootloader[J].计算机应用研究,2007,12：238～240
[16]刘文,徐磊,盛文婷.基于ARM平台的Linux内核分析与移植研究[J].现代计算机(专业版),2011,7：72～75
[17]韦斯,丁志刚,张伟宏.LINUX下UBI子系统的研究与应用[J].2010,27(10)：68
[18]郭晖,陈光.基于OpenCV的视频图像处理应用研究[J].微型机与应用,2010,21(29)：14～17
[19]徐义鑫,韦青,尚进,程德芳.Qt在嵌入式系统中的移植研究与实现[J].电子技术,2008,9(45)：49～51
[20]徐家,陈奇.基于V4L2的视频设备驱动开发[J].计算机工程与设计,201016(31)：3569-3580
[21]Ming-Hsuan Yang, David J. Kriegman, Narendra Ahuja, "Detecting Faces in Images:A Survey"[C].IEEE Transaction on Pattern Analysis and Machine Intelligence, January 2002,1(24):34-58
[22]Viola P, Jones M. Rapid object detection using a Boosted cascade of simple features [C]. Proc of IEEE Conference on Computer Vision and Pattern Recognition. Kauai, Hawaii, 2001:511-518
[23]任艳斐.直方图均衡化在图像处理中的应用[J].科技信息.2007.4：37～38
[24]李建中,雷立礼,黎灏.基于几何特征的动态人脸识别[J].科学技术与工程.2010.28：6894～6899
[25]谢毓湘,王卫威,栾悉道.基于肤色与模板匹配的人脸识别[J].计算机工程与科学.2008,30(6)：54～56
[26]余龙华,王宏,钟洪声.基于隐马尔科夫模型的人脸识别[J].计算机技术与发展.2012,22(2)：25～29
[27]刘晓东.基于弹性图匹配法的人脸识别算法研究[D].[硕士学位论文].上海：上海交通大学,2009
[28]Pentland A,Choudhury T. Face recognition for smart environments [C]. IEEE Computer, 2000,50-55
[29]孟海东,张玉英,宋飞燕.一种基于加权欧氏距离聚类方法的研究[J].计算机应用,2006,s2：152-153
[30]T.Choudhu,B.Clarkson,T.Jebara.Multimodal person recognition using unconstrained audio and video. AVBPA'99,1999,176-181
[31]R.Brunelli,D.Falavigna.Personal Identification Using MultiPle Cues.IEEE Trans.Pattern Analysis and Machine Intelligence,1995,17(10):955-966
[32]J.Li,Y.Wang,T.Tan,and A.K.Jain.Live face detection based on the analysis of fourier spectra,SPIE.2004:296-303
[33]Klaus Kollreider,Hartwig Fronthaler,Mayeel Isaae Faraj.Real-time Face Deteotion and Motion Application in "Liveness" Assessment.IEEE Transactions on Information Forensics and Security,2007,2(3):548-558
[34]刘瑞安,靳世久,张希坤等.眨眼检测与眼睛跟踪[J].计算机应用.2006,26(12)：2835～2837
[35]高军,李学伟,张建.基于模板匹配的图像配准算法[J].西安交通大学学报,2007,41(3)：307～311
[36]朱桂英,张瑞林.基于Hough变换的圆检测方法[J].计算机工程与设计,2008,29(6)：1462～1465
[37]朱俊,任明武,杨章静.基于角点检测的快速匹配算法[J].南京理工大学学报,2001,35(6)755～759