基于形状外观的运动船只识别与跟踪技术研究
|
摘要
海面运动目标检测、识别与跟踪技术在海防安全、海关管理、海上缉私以及港口船只调度等场合都有比较重要的用途。海面上来往船只碰撞的预警、报警以及识别海面运动目标的类别,这些对海上安全有着重要的应用意义。
本文主要进行海面运动船只的识别与跟踪技术的研究和应用。首先概述了海上运动目标检测和跟踪的研究与发展现状;对目前主要的运动目标识别和跟踪方法进行了简要论述;对基于形状特征匹配的识别方法进行了深入的研究和应用;利用CamShift算法对运动目标进行了跟踪。本文的主要研究工作和创新点包括以下几点:
1、对图像边缘特征检测算法进行了深入研究,提出了一种基于自适应阈值的SUSAN边缘检测算法,该算法对每个模板都计算自适应阈值,克服了传统的SUSAN算子对灰度细节丰富的图像检测效果会不够理想的情况。
2、深入研究了目标的形状特征,并用形状上下文描述目标的特征,提出了采用两级递推由粗到细的识别方式,在粗匹配阶段用形状距离来描述目标模型和船只图像整体形状上的相似性快速找出候选目标,再进行基于TPS变形模板的细匹配过程,从而实现形状最佳匹配意义上的快速准确目标识别。
3、对CamShift方法进行了深入研究并把它应用于海面船只跟踪系统中,在识别出目标船只后,利用Hue信息对船只区域建立色彩直方图,对后续帧利用该信息采用CamShift对船只区域进行跟踪,也可以进行多目标的跟踪。
Maritime objects detection、recognition and Tracking technologies are very useful in many areas, such as coastal defense、customs management、marine anti-smuggling、port scheduling and other occasions. Early warning、alarm of sea collision between vessels and identify the types of moving targets in the sea, all of these have important application.
This thesis presents a method of ship identification and tracking technology. First, it describes current development status of maritime moving target detection and tracking research ,then outlines some methods of ships identification and tracking; We carry up a deep research into the ship recognition based on the shape feature matching and track the ship by CamShift algorithm. The main research innovations and contributions are summarized as follows:
1、The paper proposes an adaptive threshold of SUSAN method. Threshold is calculated in every single SUSAN template, which makes it overcome the dissatisfactory results of traditional SUSAN operator.
2、We carry up a deep research into shape feature matching. Shape context is used to describe shape points. Then a two-stage recursive algorithm for recognition is proposed. At rough matching stage, shape distance is used to describe similarity of two targets and quickly construct a small set of candidate targets. At detail matching stage, we implement it using iteration of correspondence through shape context matching and deformation using TPS model.
3、We explore the CamShift algorithm and put it into the ship tracking system. We build the hue histogram on the identified ship region and then track it using the CamShift algorithm in the following frames. And the CamShift algorithm can be applied to track multiple ships simultaneously.
本文主要进行海面运动船只的识别与跟踪技术的研究和应用。首先概述了海上运动目标检测和跟踪的研究与发展现状;对目前主要的运动目标识别和跟踪方法进行了简要论述;对基于形状特征匹配的识别方法进行了深入的研究和应用;利用CamShift算法对运动目标进行了跟踪。本文的主要研究工作和创新点包括以下几点:
1、对图像边缘特征检测算法进行了深入研究,提出了一种基于自适应阈值的SUSAN边缘检测算法,该算法对每个模板都计算自适应阈值,克服了传统的SUSAN算子对灰度细节丰富的图像检测效果会不够理想的情况。
2、深入研究了目标的形状特征,并用形状上下文描述目标的特征,提出了采用两级递推由粗到细的识别方式,在粗匹配阶段用形状距离来描述目标模型和船只图像整体形状上的相似性快速找出候选目标,再进行基于TPS变形模板的细匹配过程,从而实现形状最佳匹配意义上的快速准确目标识别。
3、对CamShift方法进行了深入研究并把它应用于海面船只跟踪系统中,在识别出目标船只后,利用Hue信息对船只区域建立色彩直方图,对后续帧利用该信息采用CamShift对船只区域进行跟踪,也可以进行多目标的跟踪。
Maritime objects detection、recognition and Tracking technologies are very useful in many areas, such as coastal defense、customs management、marine anti-smuggling、port scheduling and other occasions. Early warning、alarm of sea collision between vessels and identify the types of moving targets in the sea, all of these have important application.
This thesis presents a method of ship identification and tracking technology. First, it describes current development status of maritime moving target detection and tracking research ,then outlines some methods of ships identification and tracking; We carry up a deep research into the ship recognition based on the shape feature matching and track the ship by CamShift algorithm. The main research innovations and contributions are summarized as follows:
1、The paper proposes an adaptive threshold of SUSAN method. Threshold is calculated in every single SUSAN template, which makes it overcome the dissatisfactory results of traditional SUSAN operator.
2、We carry up a deep research into shape feature matching. Shape context is used to describe shape points. Then a two-stage recursive algorithm for recognition is proposed. At rough matching stage, shape distance is used to describe similarity of two targets and quickly construct a small set of candidate targets. At detail matching stage, we implement it using iteration of correspondence through shape context matching and deformation using TPS model.
3、We explore the CamShift algorithm and put it into the ship tracking system. We build the hue histogram on the identified ship region and then track it using the CamShift algorithm in the following frames. And the CamShift algorithm can be applied to track multiple ships simultaneously.
引文
[1]郑南宁,计算机视觉与模式识别[M].北京:国防工业出版社,1998.
[2]马颂德,张正友.计算机视觉[M].北京:科学出版社,1998.
[3]D.H.Ballard,C.M.Brown.Computer vision,Prentice-Hall,Englewood Clifs,New Jersey,1982.
[4]T.M.Blake,The detection and tracking of small fast boats using wave radar[C].International Conference on HF Radio Systems and Techniques Publication,Guildford,UK:2000,219-244.
[5]A.A.W.Smith,M.K.Teal,P.Voles.The statistical characterization of the sea for the segmentation of maritime images[C].Conference on Video/Image Processing and Multimedia Communications,2003,489-494.
[6]P.Voles,M.Teal,J.Sanderson.Target identification in a complex maritime scene[C].IEEE Colloquium on Motion Analysis and Tracking,London,UK:1999,15/1-15/4.
[7]J.G.Sanderson,M.K.Teal,T.J.Ellis.Characterisation of a Complex Maritime Scene using Fourier Space Analysis to Identify Small Craft[J].Image Processing And Its Applications,1999,250 - 254.
[8]任明武,曹雨龙,杨静宁等.复杂条件下的船舶目标检测的图象预处理[J].计算机工程,2000,10:68-70.
[9]M.W.Ren,Y.L.Cao,Z.M.Tang,etc.One effective method for ship recognition in ship locks[C].SPIE Conf.Signal Processing,Sensor Fusion and Target Recognition.vol.3720,1999,467-472.
[10]Yulong Cao,Jingyu Yang,Mingwu Ren.Online detecting ships in lock using optical flow methodiC].Proc.SPIE Int.Soc.Opt.Eng.1999:759.
[11]周颢.复杂背景下的运动目标分割和识别的关键技术研究[D].上海:上海交通大学硕士学位论文,2004.
[12]王劲松.序列图像运动检测研究及不变矩方法在船只识别中的应用[D].合肥:中国科技大学硕士学位论文,2002.
[13]D.Cheng,H Yan.Recognition of Handwritten digits based on contour information[J]. Pattern Recognition. 1998, 31(3):235-255.
[14]H.Freeman. On the encoding of arbitrary geometric configuration[J].IEEE Trans on Electric Computer.1961,10(2):260-268.
[15] K. S. Fu. A general(syntactic- semantic) approach to picture analysis[R]. In K.S. Fu and T. Kunii. editors, Picture Engineering. 1982, SpringerVerlag:56-74.
[16] Y. Ikebe, S.Miyamoto. Shape design, representation, and restoration with splines [C]. Picture Engineering, Springer Verlag,1982:75-95
[17] J. Wu, J .Leou. New polygonal approximation schemes for object shape representation [J]. Pattern Recognition. 1993,26:471-484.
[18] Bengtsson. A, Eklundh. J. 0. Shape representation by multi- scale contour approximations[J]. IEEE Trans Pattern Anal Maeh Intell. 1991, 13(l):85-93.
[19] F.S.Cohen, Z.Huang, Z.Yang. Invariant matching and identification of curves using B-splines Curve representation[J]. IEEE Transactions on Image Proeessing 1995,4:1-10.
[20]P. Chung, C. Tsai, E.Chen, Y.Sun. Polygonal approximation using a competitive Hopfield Neural Network[C]. Pattern Recognition. 1994, 27:1505-1512.
[21] M . K. Hu. Visual Pattern Recognition by Moment Invariants[J]. IRE Transactions Information Theory, 1962,8 (2):179-187.
[22] S. Loncaric , A. P. Dhawan. A morphological signature transform for shape description[C] . Pattern Recognition. 1993,26:1.29-1.37.
[23] 李弼成,彭天强等.智能图像处理技术.北京:电子工业出版社, 2004.07
[24]Ivar. Balslev, K. P. Dcring, R. D. Eriksen. Weighted central moments in Pattern Reognition[J].Pattenr Recongition Letters, 2000, 21: 381-384.
[25] Reddy B. S, Chatterji B. N. An fft—based technique for translation, rotation and scale—invariant image registration [J]. IEEE Transactionson Image Processing, 1996, 5(8) : 1266-1271.
[26] JiZhou, Jioaying shi.A robust algorithm for feature Point matching[J].Computers&GraPhics. 2002, 26:429-436.
[27] XioalongDai, SimakaKhorram. A featuer-Based Image Registartion Algorithm Using Improved Chain-Code Rersentation Combined with Invariant Moments[C]. IEEE Trnas.on Geoscience and Remote Sensing. 1999,37(5):235-238.
[28]Mallat S.A.Theory for Multiersolution Signal Deeomposition:The Wavelet Representation[J].IEEE Trnas.PaRem Anal Mach Intell.1989,11(7):674-693.
[29]Yu Zhong,Anil K,Jain,et al.Object Tracking using Deformable Templates[J].IEEE Transactions on Pattern Analusis and Machine Intelligence.2000,22(5),198-205.
[30]Shoichi Araki,Takashi Matsuoaka,Naokazu Yokoya.Real-time tracking of multiple moving object contours in a moving camera image sequence[J].IEICE Trans.2000,E83-D(7).
[31]Paragios N,Deriche R.Geodesic active contours and level sets for the detection and tracking of moving object[J].IEEE Trans Pattern Pattern Analysis and Machine Intelligence.2000,22(3):266-280.
[32]Peterfreund N.Robust tracking of position and velocity with Kalman snakes[J].IEEE Trans Pattern Pattern Analysis and Machine Intelligence.2000,22(6):564-569.
[33]Isard M,Blake A.Contour tracking by stochastic propagation of conditional density[C].Proc European Conference on Computer Vision,Cambridge,1996,343-356.
[34]Vieren C,Cabestaing F,Postaire J.Catching Moving Object with Snake for Motion Tracking[J].Pattern Recognitiom Letters.1995,16:679-985.
[35]W.J.Rucklidge.Efficiently Locating Objects Using Hausdorff Distance[J].Internation J of Computer Vision.1997,24(3):251-270.
[36]J.Shi,C.Tomasi.Good Feature to Track[C].IEEE Conference on Computer Vision and Pattern Recognition.1994,593-600.
[37]D.Comaniciu,P.Meer.Mean Shift and Application[J].Proc.Inr'l Conf.Computer Vision.1999,1197-1203.
[38]Smith S.M,Brady J.M.SUSAN-A new approach to low level image processing[J].Intenational Journal of Computer Vision.1997,23(1):45-78.
[39]张坤华,王敬儒,张启衡.多特征复合的角点提取方法[J].中国图象图形学报,2002,7(4):319-324.
[40]章毓晋.图像工程-图像处理与分析[M].北京:清华大学出版社,1999:240.
[41]Dongxiang Zhou,Yun Hui Liu,Xuenping Cal.An efficient and robust corner detection algorithm[C].Intelligent Control and Automation.WCICA2004.2004 5:15-19,4020-4024.
[42] Yu Song , Mantian Li. Research on SUSAN Based Auto-focusing Algorithm for Optical Microscope Application[C]. International Conference on Mechatronics and Automation. 2006, 1237-1241.
[43] Marcelo M. Perez , Tim J. Dennis. An Adaptive Implementation of the SUSAN Method for Image Edge and Feature Detectionimage Processing[C].International Conference, 1997, Volume 2: 394-397.
[44] 张坤华,王敬儒,张启衡.多特征复合的角点提取方法[J].中国图象图形学报, 2002,7(4) : 319-324.
[45] Serge Belongie, Jitendra Malik. Matching with Shape Contexts[R]. IEEE Workshop on Content-based Access of Image and Video Libraries (CBAIVL'00). 2000, P20.
[46] Serge Belongie, Jitendra Malik, Jan Puzicha . Shape Matching and Object Recognition Using Shape Contexts[J]. IEEE Transactions, Pattern Analysis and Machine Intelligence, 2002,24(4):509-522.
[47] M. Bolduc, M. D. Levine. A review of biologically motivated space-variant data reduction models for robotic vision . Computer Vision and Image Understanding,1998,69(2):170-184.
[48]Serge Belongie, Jitendra Malik, Jan Puzicha . Matching Shape[C]. Computer Vision,2001, 454-461.
[49] C. Persoon, K. Stieglitz. Combinatorial Optimization:Alogorithms and Complexity. Prentice Hall, 1982.
[50] F L Bookstein. Principal warps: Thin-plate splines and the decomposition of deformation [J]. IEEE Pattern Analysis and Machine Intelligence, 1989,11(6): 567-585.
[51] Y Zheng, D Doermann. Robust point matching for nonrigid shapes by preserving local neighborhood structures [J]. IEEE Pattern Analysis and Machine Intelligence, 2006, 28(4): 643-649.
[52] H Chui, A Rangarajan. A new point matching algorithm for non-rigid registration [J]. Computer Vision and Image Understanding .2003, 89(2):114-141.
[53] S Belongie, J Malik, J Puzicha. Shape matching and object recognition using shape contexts [J]. IEEE Pattern Analysis and Machine Intelligence(S0162-8828), 2002, 24(4): 509-522.
[54] D. Comaniciu and P. Meer. Robust analysis of feature spaces: Color image segmentation[C]. International Conference on Computer Vision and Pattern Recognition. San Juan, Puerto Rico.1997, 750-755.
[55]Gray R, Bradski and Santa Clara. Computer Vision Face Tracking For Use in a perceptual User Interface[A]. Intel Technology Journal, 1998, 2: 1-15.
[56] Dorin Comaniciu, Visvanathan Ramesh, Peter Meer. Real Time Tracking of Non Rigid Objects using Mean-Shift[C]. IEEE Computer Vision and Pattern Recognition. 2000, 142-149
[57] Cheng Y. Mean shift , mode seeking, and clustering[J]. IEEE Trans. Pattern Anal. Machine Intell.1995(17):790-799.
[2]马颂德,张正友.计算机视觉[M].北京:科学出版社,1998.
[3]D.H.Ballard,C.M.Brown.Computer vision,Prentice-Hall,Englewood Clifs,New Jersey,1982.
[4]T.M.Blake,The detection and tracking of small fast boats using wave radar[C].International Conference on HF Radio Systems and Techniques Publication,Guildford,UK:2000,219-244.
[5]A.A.W.Smith,M.K.Teal,P.Voles.The statistical characterization of the sea for the segmentation of maritime images[C].Conference on Video/Image Processing and Multimedia Communications,2003,489-494.
[6]P.Voles,M.Teal,J.Sanderson.Target identification in a complex maritime scene[C].IEEE Colloquium on Motion Analysis and Tracking,London,UK:1999,15/1-15/4.
[7]J.G.Sanderson,M.K.Teal,T.J.Ellis.Characterisation of a Complex Maritime Scene using Fourier Space Analysis to Identify Small Craft[J].Image Processing And Its Applications,1999,250 - 254.
[8]任明武,曹雨龙,杨静宁等.复杂条件下的船舶目标检测的图象预处理[J].计算机工程,2000,10:68-70.
[9]M.W.Ren,Y.L.Cao,Z.M.Tang,etc.One effective method for ship recognition in ship locks[C].SPIE Conf.Signal Processing,Sensor Fusion and Target Recognition.vol.3720,1999,467-472.
[10]Yulong Cao,Jingyu Yang,Mingwu Ren.Online detecting ships in lock using optical flow methodiC].Proc.SPIE Int.Soc.Opt.Eng.1999:759.
[11]周颢.复杂背景下的运动目标分割和识别的关键技术研究[D].上海:上海交通大学硕士学位论文,2004.
[12]王劲松.序列图像运动检测研究及不变矩方法在船只识别中的应用[D].合肥:中国科技大学硕士学位论文,2002.
[13]D.Cheng,H Yan.Recognition of Handwritten digits based on contour information[J]. Pattern Recognition. 1998, 31(3):235-255.
[14]H.Freeman. On the encoding of arbitrary geometric configuration[J].IEEE Trans on Electric Computer.1961,10(2):260-268.
[15] K. S. Fu. A general(syntactic- semantic) approach to picture analysis[R]. In K.S. Fu and T. Kunii. editors, Picture Engineering. 1982, SpringerVerlag:56-74.
[16] Y. Ikebe, S.Miyamoto. Shape design, representation, and restoration with splines [C]. Picture Engineering, Springer Verlag,1982:75-95
[17] J. Wu, J .Leou. New polygonal approximation schemes for object shape representation [J]. Pattern Recognition. 1993,26:471-484.
[18] Bengtsson. A, Eklundh. J. 0. Shape representation by multi- scale contour approximations[J]. IEEE Trans Pattern Anal Maeh Intell. 1991, 13(l):85-93.
[19] F.S.Cohen, Z.Huang, Z.Yang. Invariant matching and identification of curves using B-splines Curve representation[J]. IEEE Transactions on Image Proeessing 1995,4:1-10.
[20]P. Chung, C. Tsai, E.Chen, Y.Sun. Polygonal approximation using a competitive Hopfield Neural Network[C]. Pattern Recognition. 1994, 27:1505-1512.
[21] M . K. Hu. Visual Pattern Recognition by Moment Invariants[J]. IRE Transactions Information Theory, 1962,8 (2):179-187.
[22] S. Loncaric , A. P. Dhawan. A morphological signature transform for shape description[C] . Pattern Recognition. 1993,26:1.29-1.37.
[23] 李弼成,彭天强等.智能图像处理技术.北京:电子工业出版社, 2004.07
[24]Ivar. Balslev, K. P. Dcring, R. D. Eriksen. Weighted central moments in Pattern Reognition[J].Pattenr Recongition Letters, 2000, 21: 381-384.
[25] Reddy B. S, Chatterji B. N. An fft—based technique for translation, rotation and scale—invariant image registration [J]. IEEE Transactionson Image Processing, 1996, 5(8) : 1266-1271.
[26] JiZhou, Jioaying shi.A robust algorithm for feature Point matching[J].Computers&GraPhics. 2002, 26:429-436.
[27] XioalongDai, SimakaKhorram. A featuer-Based Image Registartion Algorithm Using Improved Chain-Code Rersentation Combined with Invariant Moments[C]. IEEE Trnas.on Geoscience and Remote Sensing. 1999,37(5):235-238.
[28]Mallat S.A.Theory for Multiersolution Signal Deeomposition:The Wavelet Representation[J].IEEE Trnas.PaRem Anal Mach Intell.1989,11(7):674-693.
[29]Yu Zhong,Anil K,Jain,et al.Object Tracking using Deformable Templates[J].IEEE Transactions on Pattern Analusis and Machine Intelligence.2000,22(5),198-205.
[30]Shoichi Araki,Takashi Matsuoaka,Naokazu Yokoya.Real-time tracking of multiple moving object contours in a moving camera image sequence[J].IEICE Trans.2000,E83-D(7).
[31]Paragios N,Deriche R.Geodesic active contours and level sets for the detection and tracking of moving object[J].IEEE Trans Pattern Pattern Analysis and Machine Intelligence.2000,22(3):266-280.
[32]Peterfreund N.Robust tracking of position and velocity with Kalman snakes[J].IEEE Trans Pattern Pattern Analysis and Machine Intelligence.2000,22(6):564-569.
[33]Isard M,Blake A.Contour tracking by stochastic propagation of conditional density[C].Proc European Conference on Computer Vision,Cambridge,1996,343-356.
[34]Vieren C,Cabestaing F,Postaire J.Catching Moving Object with Snake for Motion Tracking[J].Pattern Recognitiom Letters.1995,16:679-985.
[35]W.J.Rucklidge.Efficiently Locating Objects Using Hausdorff Distance[J].Internation J of Computer Vision.1997,24(3):251-270.
[36]J.Shi,C.Tomasi.Good Feature to Track[C].IEEE Conference on Computer Vision and Pattern Recognition.1994,593-600.
[37]D.Comaniciu,P.Meer.Mean Shift and Application[J].Proc.Inr'l Conf.Computer Vision.1999,1197-1203.
[38]Smith S.M,Brady J.M.SUSAN-A new approach to low level image processing[J].Intenational Journal of Computer Vision.1997,23(1):45-78.
[39]张坤华,王敬儒,张启衡.多特征复合的角点提取方法[J].中国图象图形学报,2002,7(4):319-324.
[40]章毓晋.图像工程-图像处理与分析[M].北京:清华大学出版社,1999:240.
[41]Dongxiang Zhou,Yun Hui Liu,Xuenping Cal.An efficient and robust corner detection algorithm[C].Intelligent Control and Automation.WCICA2004.2004 5:15-19,4020-4024.
[42] Yu Song , Mantian Li. Research on SUSAN Based Auto-focusing Algorithm for Optical Microscope Application[C]. International Conference on Mechatronics and Automation. 2006, 1237-1241.
[43] Marcelo M. Perez , Tim J. Dennis. An Adaptive Implementation of the SUSAN Method for Image Edge and Feature Detectionimage Processing[C].International Conference, 1997, Volume 2: 394-397.
[44] 张坤华,王敬儒,张启衡.多特征复合的角点提取方法[J].中国图象图形学报, 2002,7(4) : 319-324.
[45] Serge Belongie, Jitendra Malik. Matching with Shape Contexts[R]. IEEE Workshop on Content-based Access of Image and Video Libraries (CBAIVL'00). 2000, P20.
[46] Serge Belongie, Jitendra Malik, Jan Puzicha . Shape Matching and Object Recognition Using Shape Contexts[J]. IEEE Transactions, Pattern Analysis and Machine Intelligence, 2002,24(4):509-522.
[47] M. Bolduc, M. D. Levine. A review of biologically motivated space-variant data reduction models for robotic vision . Computer Vision and Image Understanding,1998,69(2):170-184.
[48]Serge Belongie, Jitendra Malik, Jan Puzicha . Matching Shape[C]. Computer Vision,2001, 454-461.
[49] C. Persoon, K. Stieglitz. Combinatorial Optimization:Alogorithms and Complexity. Prentice Hall, 1982.
[50] F L Bookstein. Principal warps: Thin-plate splines and the decomposition of deformation [J]. IEEE Pattern Analysis and Machine Intelligence, 1989,11(6): 567-585.
[51] Y Zheng, D Doermann. Robust point matching for nonrigid shapes by preserving local neighborhood structures [J]. IEEE Pattern Analysis and Machine Intelligence, 2006, 28(4): 643-649.
[52] H Chui, A Rangarajan. A new point matching algorithm for non-rigid registration [J]. Computer Vision and Image Understanding .2003, 89(2):114-141.
[53] S Belongie, J Malik, J Puzicha. Shape matching and object recognition using shape contexts [J]. IEEE Pattern Analysis and Machine Intelligence(S0162-8828), 2002, 24(4): 509-522.
[54] D. Comaniciu and P. Meer. Robust analysis of feature spaces: Color image segmentation[C]. International Conference on Computer Vision and Pattern Recognition. San Juan, Puerto Rico.1997, 750-755.
[55]Gray R, Bradski and Santa Clara. Computer Vision Face Tracking For Use in a perceptual User Interface[A]. Intel Technology Journal, 1998, 2: 1-15.
[56] Dorin Comaniciu, Visvanathan Ramesh, Peter Meer. Real Time Tracking of Non Rigid Objects using Mean-Shift[C]. IEEE Computer Vision and Pattern Recognition. 2000, 142-149
[57] Cheng Y. Mean shift , mode seeking, and clustering[J]. IEEE Trans. Pattern Anal. Machine Intell.1995(17):790-799.