基于Mean Shift算法和Particle Filter算法的目标跟踪
详细信息    本馆镜像全文|  推荐本文 |  |   获取CNKI官网全文
摘要
目标跟踪技术在自动控制、监控技术、医学图像识别等领域的应用中有着独特的优势,但是近年来,目标跟踪技术仍然不能达到人们满意的效果,严重阻碍了其应用推广,随着硬件技术的飞快发展,动态图像的分析和理解成为研究的热点,目标跟踪技术应用在许多重要领域,使得该技术成为一个重要的研究课题。
     论文讨论了目前两种经典的目标跟踪算法:Mean Shift算法(均值偏移)和粒子滤波算法(Particle Filter),分析了两种算法的特点;针对目标跟踪鲁棒性不高的问题,分析了用运动目标检测提取目标运动特征的技术,通过增加对目标特征描述信息,提高跟踪健壮性,并在以颜色直方图描述颜色特征的基础上,融合了目标的运动特征,设计了一种基于运动特征和颜色特征多特征融合的粒子滤波跟踪方法。
     论文对颜色直方图进行了分析,比较了颜色直方图和二阶直方图的优劣,结合了运动特征提取方法和扩大搜索范围的方法,给出了以均值偏移为理论基础,用二阶直方图描述目标颜色特征,实现目标跟踪的技术;针对粒子滤波存在的缺点,用二阶直方图描述颜色特征,设计了均值偏移和粒子滤波相结合的目标跟踪技术,使每个粒子表示状态更合理,在遮挡时能够实现很好的跟踪。
Object tracking have special advantage in the automatic control, scout system,medical science picture identify etc. In recent years, Target tracking technique still can'tattain people's satisfaction, seriously obstructed its application expansion, so it is an urgentproblem to research target tracking. The difficult problem in visual tracking is performingfast and reliable matching of target from frame to frame
     The thesis explores currently two kinds of targets tracking algorithm, Mean Shiftalgorithm and Particle Filter algorithm. They are very good algorithm in visual targettracking area. The thesis explores the Detection of Moving Targets, To improve therobustness of visual tracking in complex environments, a novel tracking method based onadaptive fusion and particle filter is proposed, the image color and moving cues areadaptively fused to represent the target observation.
     The thesis explores an image description method based on second order histogramand increase searching scope to improves robustness of visual tracking. An algorithmbased on kernel histogram particle filter is studied, the dimension of particle is reduced andthe required particle is very few, New algorithm makes the best of middle value of particlefilter, so that the complexity of algorithm is not added.
引文
1.石华伟夏利民,基于Mean Shift算法和粒子滤波器的人眼跟踪,计算机工程与应用,2006 1-5
    2.朱胜利,Mean Shift及相关算法在视频跟踪中的研究,浙江大学,2006 14-30
    3.戴丁樟,粒子滤波算法研究及其在目标跟踪中的应用,哈尔滨工业大学,2006 12-44
    4. Yizong Cheng. Mean Shift mode Seeking and Clustering. IEEE Transactions on Pattern Analysis and Machine Intelligence. 1995.17(8) 6-10
    5. Han Yanfang and Shi Pengfei. Mean shift texture surface detection based on WT and COM feature image selection. Journal of Zhejiang University SCIENCE A iSSN 1862-1775. 2005
    6. R. T. Collins. Mean-shift blob tracking through scale space. In Proceedings of the IEEE Conference on Computer Visionand Pattern Recognition. 2003. 9-10
    7.王来雄 黄士坦,一种新的粒子滤波算法,武汉大学学报,2006 1-5
    8.莫以为 萧德云,进化粒子滤波算法及其应用,控制理论与应用,2005 3-6
    9.康健 司锡才 芮国胜,基于贝叶斯原理的粒子滤波技术概述,现代雷达,2004 1-3
    10.徐钟济,蒙特卡罗方法,上海科学技术出版社,1985
    11.胡士强 敬忠良,粒子滤波算法综述,控制与决策,2005 1-6
    12. Pan Jianxin. OUTLIERS AND INFLUENTIAL OBSERVATIONS IN A RIDGE MEAN SHIFT REGRESSION. Systems Science and Mathematical Sciences. 1995
    13. Zhaoxue Chen and Pengfei Shi. Urban road area recognition in ITS based on mean shift method. CHINESE OPTICS LETTERS. 2003
    14.熊天平,体育视频中运动目标跟踪技术研究,首都师范大学,2006 11-29
    15.师丽娜 涂峰 朱红,复杂背景下的运动目标检测方法,电子工程师,2006:(32)5-10
    16.陈朝阳 张桂林,基于图像对称差分运算的运动小目标检测方法,华中理工大学学报,1998:(26)1-3
    17.王辉,背景图像处理,哈尔滨理工大学,2005:(TP391)12-25
    18.艾海舟 王栓 何克忠,基于差分图像的人脸检测,中国图像图形学报,1998:(3)1-6
    19.危水根 陈震 黎明,一种基于时间差分运动检测的改进方法,南昌航空工业学院学报,2005:(19)1-5
    20.刘新海 方康玲,一种基于投影法的差分图像定位算法,计算机应用与软件,2004:(21)1-2
    21.胡明昊 任明武 杨静宇,一种基于直方图模式的运动目标实时跟踪算法,计算机工程与应用,2004:(TP391)1-2
    22.张昊 黄战华 郁道银,使用粒子滤波和差分进化法实现轮廓跟踪,光电工程,2004: (33) 1-4
    23.伍茜 沈季胜 刘震涛 俞小莉,动态图像差分法在热裂纹提取上的应用,兵工学报,2006:(27) 1-5
    24. Alireza Mustafa and Moteaal Asadi Shirzi and Muhammad Reza Hairi Yazdi 1. Active Tracking Using Kernel-Based Vision Processor and Robust Fuzzy Control. Control and Robotics Lab..University of Tehran.2005: (3) 2-10
    25. Anping Li, Zhongliang Jing, and Shiqiang Hu. Particle filter based visual tracking with multiy-cue adaptive fusion, CHINESE OPTICS LETTERS.2005: (3) 3-10
    26.刘芳 王涛 周登文,基于颜色-空间二维直方图的图像检索,计算机工程应用,2002:(TP391)
    27.边肇祺 张学工,模式识别,清华大学出版社,1999 9-81
    28. Stanley T.Birchfield and Sriram Rangarajan. Spatiograms Versus Histograms for Region-Based Tracking, Electrical and Computer Engineering Department.2003 1-10
    29.芦蓉,沈毅,一种改进的二维直方图的图像阀值分割方法,系统下程与电子技术,2004:(30) 1-6
    30. Dieter Fox. Adapting the Sample Size in Particle Filters Through KLD-Sampling. Department of Computer Science & Engineering University of Washington.2005 2-27
    31. Gary R. Bradski, Microcomputer Research Lab, Santa Clara, CA, Computer Vision Face Tracking For Use in a Perceptual UserInterface, Intel Corporation,2003
    32.张兆礼.现代图像处理技术及Matlab实现.第1版.北京:人民邮电出版社,2001
    33.任明武.数字图像处理.南京理工大学计算机科学与技术学院模式识别与智能控制教研室.2002
    34.盛骤等,概率论与数理统计,北京高等教育出版社,1989
    35.夏德深,现代图像处理技术与应用,南京:东南大学出版社,2001
    36.彭宁嵩,Mean-Shift可视跟踪中的稳健性研究,上海交通大学 P23-66

© 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