动态图像的自动跟踪和识别技术研究
|
摘要
在人所感知的环境信息中,视觉信息占了非常大的比重,其中动态视觉信息更是其主要组成部分。基于视频运动目标检测与跟踪融合了图像处理、模式识别、人工智能、自动控制以及计算机等许多领域的先进技术,已经成为计算机视觉研究的重要领域。目前,在复杂场景、大范围、多目标的情况下,运动目标的检测跟踪的效果不是很理想,需要进一步改善。就此现状,本文实现了在实验室环境下对运动目标进行识别和跟踪的系统,并通过控制云台、镜头将目标始终保持在视场之中。主要讨论了运动目标的识别和运动目标自动跟踪两种情况。
使用Haar特征对目标进行识别,只要训练样本充分,可以达到很好的识别效果。在识别的基础上,使用CamShift算法对目标跟踪;将Kalman滤波预测方法融入到CamShift算法中,提高了目标跟踪速度,有效地解决了有干扰情况下的目标快速跟踪问题。应用本文研究的目标自动识别与跟踪方法,建立了基于云台摄像机的快速目标自动识别与跟踪系统,并进行了一系列目标自动识别与跟踪试验,分别实现了在简单背景、复杂背景、有干扰和遮挡等多种场景下的目标自动识别与跟踪。实验结果表明:本文建立的目标自动识别与跟踪算法速度快,跟踪效果好。
The visual information, especially the dynamic visual information, occupies most of the composition among the environment information perceived by human beings.Therefore, the dynamic visual information has become an important research field of the computer vision in the perceiving environment. Object detection and tracking based on video-stream, which includes up-to-date technologies like image processing, pattern recognition, artificial intelligence,automatic control, computer science, etc, is becoming an important domain in the area of computer image processing. But currently, the detection and tracking technologies are not perfect in detecting and tracking the moving object in complex environment or in large area with multi-targets. The methods still need to be improved. According to current status,this paper realizes a system which can recognize and track moving object in laboratory, and make the target always lie in the center of imaging frame by controlling a servo devices and lens. In this paper, object’recognizing and objects’automatically tracking have been discussed.
Haar feature is used to identify the object, as long as the training samples is enough, the identification can be achieved good results。Tracking object on use of algorithm CamShift based on identification. High-speed tracking problem under disturbed situations could be solved effectively and easily by merging Kalman filter into CamShift algorithm. Applying these methods researched in this thesis, a system with CCD camera, which can detect and track object quickly and automatically, is built. The real-time and efficiency of this system is verified by some object detection and tracking experiments in different scenes such as simple background, complex background, jamming and obstruct.
使用Haar特征对目标进行识别,只要训练样本充分,可以达到很好的识别效果。在识别的基础上,使用CamShift算法对目标跟踪;将Kalman滤波预测方法融入到CamShift算法中,提高了目标跟踪速度,有效地解决了有干扰情况下的目标快速跟踪问题。应用本文研究的目标自动识别与跟踪方法,建立了基于云台摄像机的快速目标自动识别与跟踪系统,并进行了一系列目标自动识别与跟踪试验,分别实现了在简单背景、复杂背景、有干扰和遮挡等多种场景下的目标自动识别与跟踪。实验结果表明:本文建立的目标自动识别与跟踪算法速度快,跟踪效果好。
The visual information, especially the dynamic visual information, occupies most of the composition among the environment information perceived by human beings.Therefore, the dynamic visual information has become an important research field of the computer vision in the perceiving environment. Object detection and tracking based on video-stream, which includes up-to-date technologies like image processing, pattern recognition, artificial intelligence,automatic control, computer science, etc, is becoming an important domain in the area of computer image processing. But currently, the detection and tracking technologies are not perfect in detecting and tracking the moving object in complex environment or in large area with multi-targets. The methods still need to be improved. According to current status,this paper realizes a system which can recognize and track moving object in laboratory, and make the target always lie in the center of imaging frame by controlling a servo devices and lens. In this paper, object’recognizing and objects’automatically tracking have been discussed.
Haar feature is used to identify the object, as long as the training samples is enough, the identification can be achieved good results。Tracking object on use of algorithm CamShift based on identification. High-speed tracking problem under disturbed situations could be solved effectively and easily by merging Kalman filter into CamShift algorithm. Applying these methods researched in this thesis, a system with CCD camera, which can detect and track object quickly and automatically, is built. The real-time and efficiency of this system is verified by some object detection and tracking experiments in different scenes such as simple background, complex background, jamming and obstruct.
引文
[1] A.Azarbayejani, C.Wren, and A.Pentland. Real-time 3-d tracking of the human body[C].IEEE Transactions on Pattern Analysis and Machine Intelligence.Vol.19, No.7, pp 780-785, 1997
[2] Christian Schlegel, Heiko Jaberg, and Matthias Schuster.Vision Based Person Tracking with a Mobile Robot[C].British Machine Vision Conference. pp.419-427, 1997.
[3] Q.Cai, A.Mitchie, and j.K.Aggarwal. Tracking human motion in an indoor environment[C], In 2nd International Conference on Image Processing. vol. 1, pp. 215 – 218, 1995
[4] C.Wren, A.Azarbayejani, T.Darrell, and A.Pentland. Real-Time Tracking of the Human Body[C],In Proc. Of the SPIE Conference on Integration Issues in Large Commercial Media Delivery Systems, October,1995
[5] 王铨,艾海州,何克忠.基于差分图像的多运动目标的检测和跟踪[J].中国图像图形学报.1999,4(6):470-474
[6] 李振玉.图像通信与监控系统[M].北京:中国铁道出版社.1994:23-25
[7] 刘晓东,苏光大,周全,田超.一种可视化智能户外监控系统[J].中国图像图形学报.2000,5(12):1024-1029
[8] 刘振安,颜廷荣,张蕊.静态背景中的动态图像识别[J] .微机发展.2001,No.1
[9] 杨淑莹.VC++图像处理程序设计[M].北京:清华大学出版社,2005
[10] 吴立德.基于距离图象的定性识别[J].自动化学报,Vol.19,No.1,1993:63-P70
[11] 赵军.基于模型的飞机识别方法研究[D].硕士论文.西安:西北工业大学,2004
[12] 郑南宁, 计算机视觉与模式识别[M], 国防工业出版社, P1-20,1998
[13] 田捷等, 实用图象分析与处理技术[M], 电子工业出版社, P5-80,1995
[14] 蔡元龙,模式识别[M],西安电子科技大学出版社,P49-104,1990
[15] 郭军等, 发展中的文字识别理论与技术[J], 电子学报, Vol.23, No.10, 1995:184～187
[16] 吴立德.计算机视觉[M].复旦大学出版社.1993:193-200
[17] 章毓晋.图像处理和分析上册[M].清华大学出版社.1999,5:218-251
[18] 章毓晋.图像处理和分析下册[M].清华大学出版社.1999,5:190-205
[19] 夏良正,数字图像处理[M],东南大学出版社,1999,9
[20] 张广超,宋文爱.基于调整矩阵的混合噪声图像滤波[J],弹箭与制导学报
[21] 梁雯,刘松林.图像中心加权中值滤波的改进与应用[J].中国图像图形学报,1997,2(8,9):629～632
[22] 朱辉,李在铭.视频序列中运动目标检测技术[J].信号处理.2002.10.18 卷第 5 期. 449-451
[23] Viola P,Jones M. Rapid object detection using a boosted cascade of simple features[C], Proceedings of the 2001 IEEE Computer Society Conference, Computer Vision and Pattern Recognition. 2001, 1:511-518.
[24] Treptow A,Zell A. Combining Adaboost learning and evolutionary search to select features for real-time object detection[C], CEC2004, Evolutionary Computation. 2004, 2: 2107-2113
[25] 姚忠清.视频监控中的人脸检测与跟踪[D].硕士论文.成都:四川大学,2005
[26] 梁路宏,艾海舟,徐光佑等,人脸检测研究[J].计算机学报,2002, 25(5): 449-458
[27] Schapire R E. The strength of weak learnablity[M]. Machine Learning.1990. 5(2):197-227
[28] Kearns M .Vallant L G..Learning Boolean Formulae or Factoring[R]. Cambridge.MA:Havard University Aiken Computation Laboratory.1988
[29] Valiant L G. A theory of the learnable[C].Communications of the ACM .1984,27(11):1134-1142
[30] Freund Y. Boosting a weak learning algorithm by majority[M]. Information and computation, 1995, 141(2):256- 285
[31] 涂承胜,鲁明羽,陆玉昌,Boosting 家族 AdaBoost 系列代表算法[J],计算机科学,2003 Vol .30 No. 3
[32] 涂承胜,陆玉昌,Boosting 理论基础[J].计算机科学,2004 Vo1. 31 No.10
[33] 孔凡之,张兴周,谢耀菊.基于 Adaboost 的人脸检测技术[J].应用科学,2005 32(6): 7-10
[34] 赵丽红,刘纪红,徐心和.人脸检测方法综述[J].计算机应用研究,2004,21(9): 1-4
[35] 赵万鹏,古乐野.基于 Adaboost 的手写体数宇识别[J].计算机应用,2005 25(10): 2413-2415
[36] Bo Wu,Haizhou Ai, Chang Huang, Shihong Lao. Fast rotation invariant multi-view face detection based on real Adaboost [C], Sixth IEEE International Conference, Automatic Face and Gesture Recognition. 2004: 79-84
[37] 王海川,张立明.一种新的 Adaboost 快速训练算法[J].复旦学报(自然科学版),2004,43(1):27-33
[38] ZhenQiu Zhang,Mingling Li,Li. S.Z,HongJiang Zhang. Multi-view face detection with F1oatBoost[C], WACV 2002 Sixth IEEE Workshop, Applications of Computer Vision. 2002: 184-188
[39] 吴全,朱兆达.图像处理中灰度级阈值选取方法 30 年(1962-1992)的进展(一)[C].数据采集与处理.1993:193-201
[40] Soriano M,Martinkauppi B,Huovinen S.Skin detection in video under changing illumination conditions[C].Proceedings of the International Conference on Pattern Recognition,2000:839-842
[41] Nikhil R. Pal et,A Review on Image Segmentation Techniques[C],Pattern Recognition, Vol.26, No.9, 1993:1277-1294
[42] Ying D.Face-texture model based on SGLD and its application in face detection in a color scene[C].Pattern Recognition,1996,29[6]:1007-1017
[43] Peter Meer, Real-Time Tracking of Non-Rigid Objects using Mean Shift[C],IEEE CVPR,2000
[44] Cz R,Bradski.Computer Vision Face Tracking as a Component of a perceptual User Interface[J].Proc.IEEE Workshop Applications of Computer Vision,1998,(10):214-219
[45] 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[J].计算机工程与应用.2004,3
[46] Grewal, Mohinder S. and Angus P. Andrews. Kalman Filtering Theory and Practice[M].Upper Saddle River, NJ USA, Prentice Hall. 1993.
[47] Welch and Bishop. An introduction to the Kalman Filter[J]. UNC-Chapel Hill, TR 95-041,April5, 2004
[48] 丰洪才,邓华来,刘年波.用 ActiveX 控件实现对云台和镜头的控制[J]. 计算机应用研究.2003.6,22 卷第二期,27-28
[49] 张红兵等. VB6.0 环境下利用 Mscomm 控件实现串行通信[J].微计算机信息,2002.10:67-68
[50] 谭思亮,邹超群等.Visual C++串口通信工程开发实例导航[M].人民邮电出版社.2003.1:131-189
[51] J. Barron, D. Fleet, and S. Beauchemin, Performance of optical flow techniques[J]. International Journal of Computer Vision, 12(1):1994:42-77
[52] 张杨,宋文爱.运动目标图像的检测与跟踪[D].硕士论文.太原:中北大学.2005
[53] 董蕴华,魏辉.云台镜头控制系统终端解码器的设计与实现[J].河南机电高等专科学校学报.2005,3
[54] Clark F.Olson,Maximum-Likelihood Template[C],Proceedings of the IEEE Computer Society Coference on Computer Vision and Pattern Recognition, Hilton Head,South Carolina,June 2000, Vol.2,p: 52-57
[2] Christian Schlegel, Heiko Jaberg, and Matthias Schuster.Vision Based Person Tracking with a Mobile Robot[C].British Machine Vision Conference. pp.419-427, 1997.
[3] Q.Cai, A.Mitchie, and j.K.Aggarwal. Tracking human motion in an indoor environment[C], In 2nd International Conference on Image Processing. vol. 1, pp. 215 – 218, 1995
[4] C.Wren, A.Azarbayejani, T.Darrell, and A.Pentland. Real-Time Tracking of the Human Body[C],In Proc. Of the SPIE Conference on Integration Issues in Large Commercial Media Delivery Systems, October,1995
[5] 王铨,艾海州,何克忠.基于差分图像的多运动目标的检测和跟踪[J].中国图像图形学报.1999,4(6):470-474
[6] 李振玉.图像通信与监控系统[M].北京:中国铁道出版社.1994:23-25
[7] 刘晓东,苏光大,周全,田超.一种可视化智能户外监控系统[J].中国图像图形学报.2000,5(12):1024-1029
[8] 刘振安,颜廷荣,张蕊.静态背景中的动态图像识别[J] .微机发展.2001,No.1
[9] 杨淑莹.VC++图像处理程序设计[M].北京:清华大学出版社,2005
[10] 吴立德.基于距离图象的定性识别[J].自动化学报,Vol.19,No.1,1993:63-P70
[11] 赵军.基于模型的飞机识别方法研究[D].硕士论文.西安:西北工业大学,2004
[12] 郑南宁, 计算机视觉与模式识别[M], 国防工业出版社, P1-20,1998
[13] 田捷等, 实用图象分析与处理技术[M], 电子工业出版社, P5-80,1995
[14] 蔡元龙,模式识别[M],西安电子科技大学出版社,P49-104,1990
[15] 郭军等, 发展中的文字识别理论与技术[J], 电子学报, Vol.23, No.10, 1995:184～187
[16] 吴立德.计算机视觉[M].复旦大学出版社.1993:193-200
[17] 章毓晋.图像处理和分析上册[M].清华大学出版社.1999,5:218-251
[18] 章毓晋.图像处理和分析下册[M].清华大学出版社.1999,5:190-205
[19] 夏良正,数字图像处理[M],东南大学出版社,1999,9
[20] 张广超,宋文爱.基于调整矩阵的混合噪声图像滤波[J],弹箭与制导学报
[21] 梁雯,刘松林.图像中心加权中值滤波的改进与应用[J].中国图像图形学报,1997,2(8,9):629～632
[22] 朱辉,李在铭.视频序列中运动目标检测技术[J].信号处理.2002.10.18 卷第 5 期. 449-451
[23] Viola P,Jones M. Rapid object detection using a boosted cascade of simple features[C], Proceedings of the 2001 IEEE Computer Society Conference, Computer Vision and Pattern Recognition. 2001, 1:511-518.
[24] Treptow A,Zell A. Combining Adaboost learning and evolutionary search to select features for real-time object detection[C], CEC2004, Evolutionary Computation. 2004, 2: 2107-2113
[25] 姚忠清.视频监控中的人脸检测与跟踪[D].硕士论文.成都:四川大学,2005
[26] 梁路宏,艾海舟,徐光佑等,人脸检测研究[J].计算机学报,2002, 25(5): 449-458
[27] Schapire R E. The strength of weak learnablity[M]. Machine Learning.1990. 5(2):197-227
[28] Kearns M .Vallant L G..Learning Boolean Formulae or Factoring[R]. Cambridge.MA:Havard University Aiken Computation Laboratory.1988
[29] Valiant L G. A theory of the learnable[C].Communications of the ACM .1984,27(11):1134-1142
[30] Freund Y. Boosting a weak learning algorithm by majority[M]. Information and computation, 1995, 141(2):256- 285
[31] 涂承胜,鲁明羽,陆玉昌,Boosting 家族 AdaBoost 系列代表算法[J],计算机科学,2003 Vol .30 No. 3
[32] 涂承胜,陆玉昌,Boosting 理论基础[J].计算机科学,2004 Vo1. 31 No.10
[33] 孔凡之,张兴周,谢耀菊.基于 Adaboost 的人脸检测技术[J].应用科学,2005 32(6): 7-10
[34] 赵丽红,刘纪红,徐心和.人脸检测方法综述[J].计算机应用研究,2004,21(9): 1-4
[35] 赵万鹏,古乐野.基于 Adaboost 的手写体数宇识别[J].计算机应用,2005 25(10): 2413-2415
[36] Bo Wu,Haizhou Ai, Chang Huang, Shihong Lao. Fast rotation invariant multi-view face detection based on real Adaboost [C], Sixth IEEE International Conference, Automatic Face and Gesture Recognition. 2004: 79-84
[37] 王海川,张立明.一种新的 Adaboost 快速训练算法[J].复旦学报(自然科学版),2004,43(1):27-33
[38] ZhenQiu Zhang,Mingling Li,Li. S.Z,HongJiang Zhang. Multi-view face detection with F1oatBoost[C], WACV 2002 Sixth IEEE Workshop, Applications of Computer Vision. 2002: 184-188
[39] 吴全,朱兆达.图像处理中灰度级阈值选取方法 30 年(1962-1992)的进展(一)[C].数据采集与处理.1993:193-201
[40] Soriano M,Martinkauppi B,Huovinen S.Skin detection in video under changing illumination conditions[C].Proceedings of the International Conference on Pattern Recognition,2000:839-842
[41] Nikhil R. Pal et,A Review on Image Segmentation Techniques[C],Pattern Recognition, Vol.26, No.9, 1993:1277-1294
[42] Ying D.Face-texture model based on SGLD and its application in face detection in a color scene[C].Pattern Recognition,1996,29[6]:1007-1017
[43] Peter Meer, Real-Time Tracking of Non-Rigid Objects using Mean Shift[C],IEEE CVPR,2000
[44] Cz R,Bradski.Computer Vision Face Tracking as a Component of a perceptual User Interface[J].Proc.IEEE Workshop Applications of Computer Vision,1998,(10):214-219
[45] 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[J].计算机工程与应用.2004,3
[46] Grewal, Mohinder S. and Angus P. Andrews. Kalman Filtering Theory and Practice[M].Upper Saddle River, NJ USA, Prentice Hall. 1993.
[47] Welch and Bishop. An introduction to the Kalman Filter[J]. UNC-Chapel Hill, TR 95-041,April5, 2004
[48] 丰洪才,邓华来,刘年波.用 ActiveX 控件实现对云台和镜头的控制[J]. 计算机应用研究.2003.6,22 卷第二期,27-28
[49] 张红兵等. VB6.0 环境下利用 Mscomm 控件实现串行通信[J].微计算机信息,2002.10:67-68
[50] 谭思亮,邹超群等.Visual C++串口通信工程开发实例导航[M].人民邮电出版社.2003.1:131-189
[51] J. Barron, D. Fleet, and S. Beauchemin, Performance of optical flow techniques[J]. International Journal of Computer Vision, 12(1):1994:42-77
[52] 张杨,宋文爱.运动目标图像的检测与跟踪[D].硕士论文.太原:中北大学.2005
[53] 董蕴华,魏辉.云台镜头控制系统终端解码器的设计与实现[J].河南机电高等专科学校学报.2005,3
[54] Clark F.Olson,Maximum-Likelihood Template[C],Proceedings of the IEEE Computer Society Coference on Computer Vision and Pattern Recognition, Hilton Head,South Carolina,June 2000, Vol.2,p: 52-57
© 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 |