基于FPGA机器视觉的运动目标检测跟踪系统
详细信息    本馆镜像全文|  推荐本文 |  |   获取CNKI官网全文
摘要
随着计算机技术的飞速发展,数字图像技术在工业生产、安防监控、消费电子、智能交通等领域有了广泛的应用。目标跟踪即是在此基础上发展起来的一项新技术,用以实现人工智能机器视觉,在实现视觉机械手、流水线工件分拣、缺陷产品检测、车辆跟踪等领域有较大的应用前景。
     目前运动目标跟踪因其数据量大、算法复杂、稳定性要求高等因素而成为一个难题,目前一般由ARM、DSP或PC机实现,随着各种视频质量的不断提高以及压缩格式的不断改进,对图像处理系统运算速度的要求也不断提高,其实时性、并行性、稳定性都很难满足,而FPGA由于其速度快、开发难度低、可重构性好、成本低等优势,将逐步展现优势,同台竞技。
     本研究采用CMOS摄像头高速采集图像信息,利用Altera FPGA的NiosII软核对图像进行运动目标跟踪算法处理,得到目标的大小、位置、轮廓等信息,驱动两自由度伺服舵机云台,使摄像头对目标进行跟踪,同时可实现VGA同步显示。
     运动目标跟踪系统包括摄像头驱动模块、视频采集模块、视频输入缓冲FIFO、SOPC系统、舵机PWM控制模块、视频输出缓冲FIFO、VGA显示接口模块。摄像头驱动模块采用Verilog HDL语言实现SCCB总线时序,对CMOS摄像头进行初始化配置。视频采集模块同样采用Verilog HDL语言对视频格式进行转换,将串行的8位RGB565数据转换为并行的16位RGB565数据,同时产生数据有效信号。输入输出缓冲FIFO采用FPGA自带IP核生成,用以异步时钟的同步。VGA模块用以将SDRAM中的图像实时显示到VGA显示器。PWM模块负责将运算得到的目标信息驱动到舵机云台。SOPC系统主要包括NiosII处理器、Avalon总线、DMA控制器、SDRAM控制器、自定义FIFO控制器。其中FIFO到SDRAM的数据传输采用DMA模式,以减少NiosII处理器负荷。
     系统架构在QuartusII中设计完成,系统算法在NiosII IDE集成开发环境中用C语言进行设计和调试,主要包括了系统底层驱动程序开发与上层应用程序开发。最后使用QuartusII对整个系统工程经行综合与布局布线,通过Modelsim进行功能与时序仿真。在时序分析报告满足要求后将配置文件pof或sof下载到FPGA中。
     实验结果表明,所设计的系统能满足预期要求,在实时性与准确性等方面都有较好的表现,为后期的进一步发展打下了基础。
With the development of computer technique, digital video technology has a widerange of application in these fields such as industrial production、security monitoring、consumer electronics、intelligent transportation and so on. And on that basis, a newtechnology-Target tracking is developed which is used to realize artificial intelligencemachine vision and has a larger application prospect in the domains like visionmanipulator achieved、line work piece sorted、defect product testing、vehicle tracking.
     At present, moving target tracking technology implemented with ARM﹑DSP orPC becomes a problem for its large amount of data、complex algorithm and highstability. Along with all kinds of high quality video developed and compressed formatcontinuously improved, operation speed of image processing system becomes higherand higher and its real-time、parallel、stability is difficult to meet. However, FPGA willgradually show the advantages such as high speed、low developing difficulty、wellreconstruction and low cost, and compete with others.
     In the study, the CMOS camera is adopted to high-speed collect image informationand NiosII soft IP core of Altera FPGA is used to process image which can get the size、location and the information of the target, and then drives servo steering gear tripodhead to make the camera to track the target and also can realize VGA displayed.
     Moving target tracking system includes camera driving module、video acquisitionmodule、the video input buffer FIFO、SOPC system、steering gear PWM controlmodule、video output buffer FIFO、VGA display interface module. Camera drivingmodule uses Verilog HDL language to implement SCCB bus timing and configurateCMOS camera initially. Video acquisition module also uses Verilog HDL language toconvert video format, which will transform a serial of8bit RGB565data into a parallel16bit RGB565data, and produce data effective signal. Input/output buffer FIFO isbased on FPGA own IP cole generation to make asynchronous clock to synchronize.VGA module is used to make the image of SDRAM to display to VGA displayer onreal-time. PWM module is responsible for driving the target information by calculatedto steering gear tripod head. The SOPC system includes NiosII processor, Avalon bus,DMA controller, SDRAM controller, and custom FIFO controller. Among them thedata transmitted from FIFO to SDRAM is adopted by DMA mode, in order to reducethe load of NiosII processor.
     The architecture of system finishes in QuartusII, the system algorithm designs anddebugs in NiosII IDE integrated development environment using C language, whichmainly includes the system bottom driver program and the upper applicationdevelopment. Finally, through the Modelsim function and timing simulation, QuartusIIis used to the line synthesis and locating and wiring to the whole system engineering.The configuration files are downloaded to FPGA when the sequential analysis reportmeet the requirements.
     The experimental result shows that the design of the system meets expected requirement, and has the good performance in real time and accuracy, and lay thefoundation for the further development.
引文
[1]康磊.视频监控系统原理及实现[D].北京:北京邮电大学,2010
    [2]徐正平.基于CMOS摄像头与FPGA的位置检测系统设计[J].液晶与显示,2009,5(12)
    [3]黄健群.数字式电视跟踪器与跟踪性能检测评估系统[J].电光与控制,2005,2(5)
    [4] Li Kaiyu,Zhang Huanchun,Jing Yazhi.FPGA based fast,high—quality rotationof image[J].Journal of Image and Graphics,2004,9(3)
    [5] Brown, M.; Lowe, D.G. Unsupervised3D object recognition and reconstruction inunordered datasets.3-D Digital Imaging and Modeling,2005.3DIM2005. FifthInternational Conference on13-16June2005
    [6]吕鹏.基于FPGA的数字视频侦察监控系统设计[D].南京:南京理工大学,2009
    [7]王四伟.基于SOPC的指纹采集与处理系统的研究[D].武汉理工大学,2009
    [8]谢燕平.基于FPGA的智能仪器远程控制系统设计[D].武汉:北京交通大学,2010
    [9] Mohd-Yasin, F.; Tan, A.L.; Reaz, M.I.The FPGA prototyping of iris recognitionfor biometric identification employing neural network Microelectronics,2004.ICM2004Proceedings,2004,10
    [10] Liu L.Fan G.A new JPEG2000region—of—interest image coding method:partialsignificant bitplanes shift[J]. IEEE Signal Processing Letters,2003,10(2)
    [11] Mahlknecht, Stefan, Oberhammer, Roland,Novak, Gregor. A real-time imagerecognition system for tiny autonomous mobile robots.Proceedings-IEEEReal-Time and Embedded Technology and Applications Symposium, Proceedings-RTAS200410th IEEE Real-Time and Embedded Technology and Applications Symposium,2004,v10:324-330
    [12] T.Horprasert,D.Harwood,and L.S.Davis,“A Statistical Approach for Real-timeRobust Background Subtraction and Shadow Detection,”Proc.IEEE ICCV’99FRAME-RATE Workshop,Kerkyra,Greece,1999
    [13]马慧.基于Avalon总线的VGA控制器IP核设计[J].现代计算机,2010,4(10)
    [14]田秀伟.基于SOPC的LED显示屏控制器设计[J].液晶与显示,2007,6(12)
    [15]李秉.基于多源低空信息的巡航导弹预警研究[J].雷达与电子战,2008,4(13)
    [16]刘中华.图像中多运动目标的识别和跟踪[D].上海:上海交通大学,2003
    [17]何骏.车辆智能监控管理系统设计[J].科技信息,2010,4(6)
    [18] RUI-HONG YANG, QUAN PAN, YONG-MEI CHENG. The Application of Wavelet InvariantMoments To Image Recognition. Proceedings of the Fifth International Conferenceon Machine Learning and Cybernetics, Dalian, August2006
    [19]满征瑞.交通场景中智能多目标跟踪技术研究[D].武汉:华中科技大学,2009
    [20] Sadgal, Mohammed; El Fazziki; Aziz Ouahman; Abdellah Ait. Aerial image processingand object recognition.Visual Computer,2005.2,v21:118-123
    [21]刘会杰.计算机视觉的理论框架研究[J].计算机科学,1999,3(1)
    [22]李巍.一种基于光流法的三维互动系统的研究与实现[D].武汉:华中科技大学,2007
    [23]陈伟宏.基于多摄像头的目标跟踪算法研究[D].长沙:湖南大学,2005
    [24] VAILAYA A,FIGUEIRED0M,JAIN A K.Image Classification for Content-basedIndexing[J].IEEE Transactions on Image Processing,2001,10(1)
    [25]侯志强.视觉跟踪技术综述[J].自动化学报,2006,4(12)
    [26]梁敏.基于粒子滤波的多目标跟踪算法研究[D].西安:西安电子科技大学,2010
    [27]李圆双.基于机器视觉的智能车跟踪系统研究与实现[D].武汉:中南民族大学,2011
    [28] J. Imrek. Development of an FPGA-Based Data Acquisition Module for Small AnimalPET. IEEE TRANSACTIONS ON NUCLEAR SCIENCE. OCTOBER2006. VOL.53, NO.5
    [29] Migue1Arias Estrada,Roberto Garcia de Campo.An FPGA Motion ComputationArchitecture for Real Time Performance[C].In:Visual2000Conference,Mexico,DF,2000.09
    [30]尹俊超.基于OpenCV的运动目标检测与跟踪[J].计算机工程与设计,2011,8(32)
    [31]郭强.智能交通中FPGA图像采集系统及灯控系统的研究[D].北京:北京工业大学,2008
    [32]徐正平.基于CMOS摄像头与FPGA的位置检测系统设计[J].液晶与显示,2009,5(16)
    [33]吕鹏.基于FPGA的数字视频侦察监控系统设计[D].南京:南京理工大学,2009
    [34]熊文彬.基于FPGA实现的视频显示系统[J].液晶与显示,2011,1(26)
    [35]陈志星.基于NiosⅡ的视频采集系统设计[D].武汉:武汉理工大学,2010
    [36] Wu Zhilu,Ren Guanghui,Zhao Yaqin.A Study On Implementing Wavelet Transformand FFT with FPGA[J].Proceedings.4th International Conference,2001.1
    [37]刘军.基于嵌入式系统的运动目标跟踪视频监控系统的研究[D].山东:山东大学,2011
    [38]金克琼.视频监控系统中运动目标检测与跟踪的研究[D].武汉:武汉理工大学,2010
    [39]郭文强.基于卡尔曼预测算法的云台三维空间目标跟踪[J].陕西科技大学学报,2007,1(3)
    [40] Vu K,Hua KA,Tavanapong W.Image retrieval based onregions of interest[J].IEEETransactions on Knowledge and Data Engineering,2003,15(4)
    [41] J.C.S.J.Jr., C.R.Jung and S.R.Musse,“A background subtraction model adaptedto illumination changes”, IEEE International Conference on image processing,pp.1817-1820,2006
    [42] T.Kanade, R.Collins, A.Lipton, P.Burt, and L.Wixson,“Advances in cooperativemulti-sensor video surveillance,”in Proc.DARPA Image Understanding Workshop,vol.1, pp.3–24, Nov.1998
    [43] Altera Corporation. CyeloneII Device Family DataSheet.2005.
    [44]董文明.基于背景重构的运动目标检测算法[J].重庆邮电大学学报,2008,6(16)
    [45]徐以美.视频序列中运动目标检测与跟踪技术研究[D].西安:西安电子科技大学,2008
    [46]李小红.基于DSP图像处理实验系统的开发与应用[J].合肥学院学报,2009,4(22)
    [47]李振宇.探析数字图像处理方法和技术[J].电脑编程技巧与维护,2010,24(32)
    [48]于新兴.图像数据挖掘的模型和技术[J].中国组织工程研究与临床康复,2010,35(14)
    [49]李劲菊.一种复杂背景下运动目标检测与跟踪方法[J].仪器仪表学报,2010,21(11)
    [50] R.Cucchiara,C.Grana,and A.Prati,“Detection moving object,ghosts,and shadowsin video streams,”IEEE transaction on pattern analysis and machine intelligence,vol.25,October.2003
    [51]邵娜.视频序列中运动目标检测与跟踪方法研究[J].光机电信息,2009,2(1)
    [52]吴银玲.基于运动检测与跟踪的人体目标的统计研究[D].天津:天津大学,2009
    [53]吴晖.动态场景中的目标检测算法的研究[D].合肥:合肥工业大学,2011
    [54]郑锦.视频序列中运动对象检测技术的研究现状与展望计算机应用研究[J].计算机应用研究,2008,12(7)
    [55]邵文坤.动态场景下的运动目标跟踪方法研究[J].计算机仿真,2006,5(20)
    [56]安博.动态背景下运动目标检测的研究[D].天津:天津理工大学,2009
    [57]李巍.一种基于光流法的三维互动系统的研究与实现[D].武汉:华中科技大学,2009
    [58]刘川辉.视频图像目标检测技术方法研究[D].成都:成都理工大学,2010
    [59]楚勇.变背景的运动目标检测系统[D].西安:西安理工大学,2009
    [60]许文武.基于视频的运动目标检测与跟踪方法应用研究[D].扬州:江苏科技大学,2011
    [61]舒文.实时场景下的运动目标检测技术研究[D].重庆:西南交通大学,2010
    [62]晏伯武. EDA技术及其教学相关问题的探讨[J].黄石理工学院学报,2010,1(11)
    [63]黄冰峰.基于SoPC目标板Flash编程设计的创建及应用[J].电子设计工程,2009,3(15)
    [64] Mae,Y,et al.,“Optical Flow Based Real-time object tracking by active visionsystem,”Proc.2nd Japan-France Congress on Mechatronics,1994
    [65] Tieu K,Viola R. Boosting image retrieval[J].International Journal of ComputerVision,2004,56(l/2)
    [66]陈一骄.基于FPGA的可重构硬件实现技术研究[J].信息工程大学学报,2009,1(4)
    [67]熊兴中. SDRAM内存条控制器的设计与实现[J].四川理工学院学报,2006,6(25)
    [68]郑佃好.基于ADV7125的嵌入式系统VGA接口设计[J].单片机与嵌入式系统应用,2011,8(11)
    [69]张婷婷.基于ARM和CPLD的四足机器人嵌入式控制器硬件平台设计[D].武汉:武汉科技大学,2009
    [70]张华儒.四腿机器人的运动控制系统设计[J].机械制造与自动化,2010,6(40)
    [71] Altera Corporation. NiosII Processor Reference Handbook [DB/OL]. http://www.altera.com
    [72]杨鉴英.基于SOPC的嵌入式系统设计[D].西安:西安电子科技大学,2007
    [73]徐锋.嵌入式智能家居控制器研究及家居网络开发[D].大连:大连理工大学,2008
    [74]吕涛.数字图像采集与高速传输部件的研究与设计[D].重庆:西南科技大学,2005
    [75]宋海吒.基于FPGA和OV7620的图像采集及VGA显示[J].电视技术,2011,5(35)
    [76]潘明.基于FPGA的直流电机脉宽调制控制[J].广西科学院学报,2005,4(16)
    [77] Altera Corporation. Using Signal TaPII Embedded Logic Analyzers in SoPC BuilderSystem [DB/OL]. httP://www.altera.eom.
    [78]张晓健.基于NiosⅡ的嵌入式逻辑分析仪系统设计[D].西安:西安电子科技大学,2007
    [79] Altera Corporation. NiosII Software DeveloPer handbook,[DB/OL]. http://www.altera.com
    [80]谭华. Avalon片上总线协议的形式化建模与模型检验分析[J].桂林电子科技大学学报,2009,2(25)
    [81]覃洪英.基于NiosⅡ的DMA传输及调试[J].石油仪器,2010,3(24)
    [82]杨进. SOPC中自定义FIFO接口与DMA数据传输[J].单片机与嵌入式系统应用,2008,8(12)
    [83]曾振兴.视频采集中DMA控制的流模式接口设计[J].单片机与嵌入式系统应用,2009,5(21)

© 2004-2018 中国地质图书馆版权所有 京ICP备05064691号 京公网安备11010802017129号

地址:北京市海淀区学院路29号 邮编:100083

电话:办公室:(+86 10)66554848;文献借阅、咨询服务、科技查新:66554700