基于贝叶斯背景模型的遗留物检测算法研究
|
摘要
在计算机视觉、视频监控领域,背景模型构建是一项关键技术。背景建模的结果将会对图像序列的运动目标检测、分类、跟踪及行为理解等后续处理工作产生重要影响。背景建模是序列图像分析的基础性工作,也是当今国内外学者研究的热点问题。本文对复杂场景的背景建模进行了较为深入的研究,并着重研究了贝叶斯背景估计算法。本文的主要工作如下:
首先,利用混合高斯模型算法分别对静态背景、动态背景进行背景建模。通过对实验结果的分析得出,此算法仅对静态背景或较简单的动态背景有很好的效果;然而在处理室外动态复杂场景时,构建出来的背景效果并不理想。
然后,主要针对以上混合高斯模型的不足,引入了贝叶斯理论。通过贝叶斯理论对混合高斯模型进行了多方面的改进。第一,针对高斯模型的收敛问题,应用EM算法对其进行了改进,保证算法的快速收敛。第二,在背景分割阈值选取上,根据影响阈值产生变化的因素,提出了由样本差和均值构造的分割阈值选取方法。针对背景分割图中噪声点较多和前景目标连通性较差的问题,在视频分割图像后处理阶段,分别采取连通域处理、噪声抑制和形态学技术等方法进行处理,降低了目标检测的虚警率。
最后,把上面模型估计出来的背景应用到滞留物和偷窃物的检测方面。由贝叶斯模型更新速率不同而得到的两个背景,通过背景差法得到两幅二值化的前景帧。然后再对前景帧中的运动信息分别进行累计获得特征图像,最后通过特征图像把图像分为运动物、背景和可疑目标。整个过程不需要对目标进行跟踪,便于实现。
实验结果验证了本文算法的有效性和可靠性。
In computer vision and video surveillance field, background modeling is a key technology. The higher level projects such as moving objects detection、classification、tracking and behavior understanding depend on the results of background modeling. Background modeling is one of the fundamental parts for video analysis, and which is currently a hot topic widely researched around the world. This paper focuses on the background modeling of complex scenes, and places emphasis on Bayesian background estimation algorithm. The main contributions of this paper are summarized as follows:
First, Gaussian mixture model(GMM) is used to model background images which are static and dynamic scenes, and through the analysis of experiment results, we conclude that GMM can get a good effect only when the background is static or partly dynamic;however,while dealing with complex outdoor scenes, results are not nice.
Then, Bayesian network algorithm is introduced to improve GMM in this paper,and improvements are proposed for this method. Firstly, for convergence problem of this model, the EM algorithm is applied to ensure fast convergence of the algorithm. Secondly, a thresholding method based on sample mean and standard deviation is presented, which can class pixels more accurately. At last, some measures have been taken in image post-processing such as noise suppression and Mathematical Morphology technology to remove noise and increase connectivity.
Finally, this model is applied to detect abandoned and removed objects. By processing the input video at different frame rates, two backgrounds are constructed: one for short-term and another for long-term. Two binary foreground maps are estimated by comparing the current frame with the backgrounds, and motion statistics are aggregated in a likelihood image by applying a set of foreground maps. Likelihood image is then used to differentiate between the pixels that belong to moving objects, temporarily static regions and scene background. The temporary static regions indicate abandoned items, illegally parked vehicles, objects removed from the scene. And this presented pixel-wise method does not require object tracking and can be performed easily.
The experimental results show that the approach proposed in this thesis is effective and reliable.
首先,利用混合高斯模型算法分别对静态背景、动态背景进行背景建模。通过对实验结果的分析得出,此算法仅对静态背景或较简单的动态背景有很好的效果;然而在处理室外动态复杂场景时,构建出来的背景效果并不理想。
然后,主要针对以上混合高斯模型的不足,引入了贝叶斯理论。通过贝叶斯理论对混合高斯模型进行了多方面的改进。第一,针对高斯模型的收敛问题,应用EM算法对其进行了改进,保证算法的快速收敛。第二,在背景分割阈值选取上,根据影响阈值产生变化的因素,提出了由样本差和均值构造的分割阈值选取方法。针对背景分割图中噪声点较多和前景目标连通性较差的问题,在视频分割图像后处理阶段,分别采取连通域处理、噪声抑制和形态学技术等方法进行处理,降低了目标检测的虚警率。
最后,把上面模型估计出来的背景应用到滞留物和偷窃物的检测方面。由贝叶斯模型更新速率不同而得到的两个背景,通过背景差法得到两幅二值化的前景帧。然后再对前景帧中的运动信息分别进行累计获得特征图像,最后通过特征图像把图像分为运动物、背景和可疑目标。整个过程不需要对目标进行跟踪,便于实现。
实验结果验证了本文算法的有效性和可靠性。
In computer vision and video surveillance field, background modeling is a key technology. The higher level projects such as moving objects detection、classification、tracking and behavior understanding depend on the results of background modeling. Background modeling is one of the fundamental parts for video analysis, and which is currently a hot topic widely researched around the world. This paper focuses on the background modeling of complex scenes, and places emphasis on Bayesian background estimation algorithm. The main contributions of this paper are summarized as follows:
First, Gaussian mixture model(GMM) is used to model background images which are static and dynamic scenes, and through the analysis of experiment results, we conclude that GMM can get a good effect only when the background is static or partly dynamic;however,while dealing with complex outdoor scenes, results are not nice.
Then, Bayesian network algorithm is introduced to improve GMM in this paper,and improvements are proposed for this method. Firstly, for convergence problem of this model, the EM algorithm is applied to ensure fast convergence of the algorithm. Secondly, a thresholding method based on sample mean and standard deviation is presented, which can class pixels more accurately. At last, some measures have been taken in image post-processing such as noise suppression and Mathematical Morphology technology to remove noise and increase connectivity.
Finally, this model is applied to detect abandoned and removed objects. By processing the input video at different frame rates, two backgrounds are constructed: one for short-term and another for long-term. Two binary foreground maps are estimated by comparing the current frame with the backgrounds, and motion statistics are aggregated in a likelihood image by applying a set of foreground maps. Likelihood image is then used to differentiate between the pixels that belong to moving objects, temporarily static regions and scene background. The temporary static regions indicate abandoned items, illegally parked vehicles, objects removed from the scene. And this presented pixel-wise method does not require object tracking and can be performed easily.
The experimental results show that the approach proposed in this thesis is effective and reliable.
引文
[1] T. Collins, AJ. Lipton, T. Kanade, H. Fujiyoshi, etc.A System for Video Surveillance and Monitoring [R], VSAM Final Report, Technical report CMU-RI-TR,0~12.
[2] I. Haritaoglu, D. Harwood, L. Davis.Real-Time Surveillance of People and Their Activities [C] . IEEE Transactions on Pattern Analysis and Machine Intelligence, 2000, 22(8), 809~830.
[3] R. S. Feris, Y. Tian and A. Hampapur. Capturing People in Surveillance Video [C].IEEE International Workshop on Visual Surveillance, 2007.
[4] M. P Dubuisson, A. Jaill.Contour Extraction of Moving Objects in Complex Outdoor Scenes[J].International Journal of Computer Vision, 1995, 14(1), 83~105.
[5] B.K. Horn, BQ Schunck.Determining Optical Flow [J].Artificial Intelligence, 1981, 17(1), 185~203.
[6] Y. Mac, Y. Shkm, J. Miura, etc.Object Tracking in Cluttered Background Based on Optical Flow and Edges [C].Proceedings of the Thirteenth International Conference On Pattern Recognition, 1996, 14(1), 25~29.
[7]王亮,胡卫明,谭铁牛.基于步态的身份识别[J].计算机学报,2003,26(3):353~360.
[8]田原,谭铁牛,孙洪赞.一种具有良好鲁棒性的实时跟踪方法[J].自动化学报,2002,28(5),851~853.
[9] Simon Haykin.Adaptive Filter Theory. 3rd edition [M].Prentice Hall, 1996.
[10]谭博怡,曾慧,胡占义.一种车轮轮廓的自动提取方法[J].自动化学报, 2008, 34(5), 516~522.
[11]刘蓉,段福庆,刘三阳,等.基于均值漂移的自适应滤波及其在光谱信号处理中的应用[J].电子与信息学报,2006,28,312~316.
[12]李庆忠,何东晓.基于聚类的背景建模与运动目标检测方法[J].计算机工程与应用,2008,44(8),193~195.
[13]D. Forsyth, J. Ponce,林学阖,王宏等译.计算机视觉——一种现代方法[M].北京:电子工业出版社,2004.
[14]Q. Li, D. He, B. Wang.Effective Moving Objects Detection Based on Clustering Background Model for Video Surveillance[C].IEEE International Congress on Image and Signal Processing, 2008, 03, 656~660.
[15]M. Boninsegna, A. Bozzoli.A Tunable algorithm to update a reference image [C].Signal Processing, Image Communication, 2000, 16(4), 353~36.
[16]A. Mittal, N. Paragios.Motion-based background subtraction using adaptive kernel density estimation [C] . in Proc. IEEE Conf. on Computer Vision and Pattern Recognition, 2004, 15(2) , 302~309.
[17]边肇祺等.模式识别[M].北京:清华大学出版社,2000.
[18]Han Bohyung, Comaniciu Dorin, Davis Larry.Sequential Kernel Density Approximation through Mode Propagation [C] .Applications to Background Modeling, Proceeding of the Asian Conference on Computer Vision, 2004, 03, 246~253.
[19]Katkovnik Vladimir,Shmulevich Ilya.Nonparametric Density Estimation with Adaptive Varying Window Size [C].Proceedings of SPIE, 2001, 273~281.
[20]A. Elgammal, R. Duraiswami, D. Harwood, etc.Background and foreground modeling using nonparametric kernel density estimation for visual surveillance [C].Proceedings of the IEEE, 2002, 138~146.
[21]C. Wren, A. Azarbayejani, T. Darrell, etc.Pfmder:Real-time Tracking of the Human Body [C].IEEE Transactions on Pattern Analysis and Machine Intelligence, 1997, 19(7), 780~785.
[22]C. Stauffer, E. Grimson.Adaptive background mixture models for real-time tracking [C], in Proc. IEEE Conf. on Computer Vision and Pattern Recognition, Fort Collins, CO, vol. 02, 1999, 246~252.
[23]P. Kaew, Tra Pong,R. Bowden.An Improved Adaptive Background Mixture Model for Real-time Tracking with Shadow Detection [C].Proceedings of Second European Workshop on Advanced Video Based Surveillance Systems, 2001, 149~153.
[24]Gelman Andrew, Carlin John, S. Stern, and etc.Bayesian Data Analysis 2nd Edition [M], America, chapman & halu , 2003.
[25]Oncel Tuzel, Fatih Porikli, Peter Meer.A Bayesian Approach to Background Modeling [C].CVPR'05, 03, 58 ~63.
[26]J.Q. Shi, R. Murray-Smith, D.M. Titterington.Bayesian Regression and Classification Using Mixtures of Gaussian Processes [J].Adapt. Control Signal Process, 2003, 17, 1~16.
[27]Baldi Pierre, Long. D. Anthony.A Bayesian framework for the analysis of microarray expression data: regularized t-test and statistical inferences of gene changes [J].Bioinformatics, 17, 509~519.
[28]苏建.基于EM算法的基因芯片图像信号检测与分析[D],苏州大学硕士学位论文,2009.
[29]温津伟,罗四维,赵嘉利等.基于Bayesian的期望最大化方法BEM算法[J].计算机研究与发展,2001,38(7),821~824.
[30]俞奎,王浩,吴信东,等.贝叶斯网络的并行EM学习算法[J].模式识别与人工智能,2008,21(5),670~676.
[31]K. Smith, P. Quelhas, D. Gatica-Perez.Detecting abandoned luggage items in a public space [C].in PETS, 2006, 75~82.
[32]C. Sacchi, C. Regazzoni.A distributed surveillance system for detection of abandoned objects in unmanned railway environments [C].IEEE Transactions on Vehicular Technology, 2000, 49.
[33]P.T. Krahnstoever, T. Sebastian, A. Perera, etc.Multi-view detection and tracking of travelers and luggage in mass transit environments [C].VS2006, 2006, 67~74.
[34]Tian Yingli, Feris Rogerio, Hampapur Arun.Real-Time Detection of Abandoned and Removed Objects in Complex Environments [C].VS2008, Marseille, 2008.
[35]高大山.智能交通系统中运动目标检测方法的研究[D],清华大学硕士学位论文,2003.
[2] I. Haritaoglu, D. Harwood, L. Davis.Real-Time Surveillance of People and Their Activities [C] . IEEE Transactions on Pattern Analysis and Machine Intelligence, 2000, 22(8), 809~830.
[3] R. S. Feris, Y. Tian and A. Hampapur. Capturing People in Surveillance Video [C].IEEE International Workshop on Visual Surveillance, 2007.
[4] M. P Dubuisson, A. Jaill.Contour Extraction of Moving Objects in Complex Outdoor Scenes[J].International Journal of Computer Vision, 1995, 14(1), 83~105.
[5] B.K. Horn, BQ Schunck.Determining Optical Flow [J].Artificial Intelligence, 1981, 17(1), 185~203.
[6] Y. Mac, Y. Shkm, J. Miura, etc.Object Tracking in Cluttered Background Based on Optical Flow and Edges [C].Proceedings of the Thirteenth International Conference On Pattern Recognition, 1996, 14(1), 25~29.
[7]王亮,胡卫明,谭铁牛.基于步态的身份识别[J].计算机学报,2003,26(3):353~360.
[8]田原,谭铁牛,孙洪赞.一种具有良好鲁棒性的实时跟踪方法[J].自动化学报,2002,28(5),851~853.
[9] Simon Haykin.Adaptive Filter Theory. 3rd edition [M].Prentice Hall, 1996.
[10]谭博怡,曾慧,胡占义.一种车轮轮廓的自动提取方法[J].自动化学报, 2008, 34(5), 516~522.
[11]刘蓉,段福庆,刘三阳,等.基于均值漂移的自适应滤波及其在光谱信号处理中的应用[J].电子与信息学报,2006,28,312~316.
[12]李庆忠,何东晓.基于聚类的背景建模与运动目标检测方法[J].计算机工程与应用,2008,44(8),193~195.
[13]D. Forsyth, J. Ponce,林学阖,王宏等译.计算机视觉——一种现代方法[M].北京:电子工业出版社,2004.
[14]Q. Li, D. He, B. Wang.Effective Moving Objects Detection Based on Clustering Background Model for Video Surveillance[C].IEEE International Congress on Image and Signal Processing, 2008, 03, 656~660.
[15]M. Boninsegna, A. Bozzoli.A Tunable algorithm to update a reference image [C].Signal Processing, Image Communication, 2000, 16(4), 353~36.
[16]A. Mittal, N. Paragios.Motion-based background subtraction using adaptive kernel density estimation [C] . in Proc. IEEE Conf. on Computer Vision and Pattern Recognition, 2004, 15(2) , 302~309.
[17]边肇祺等.模式识别[M].北京:清华大学出版社,2000.
[18]Han Bohyung, Comaniciu Dorin, Davis Larry.Sequential Kernel Density Approximation through Mode Propagation [C] .Applications to Background Modeling, Proceeding of the Asian Conference on Computer Vision, 2004, 03, 246~253.
[19]Katkovnik Vladimir,Shmulevich Ilya.Nonparametric Density Estimation with Adaptive Varying Window Size [C].Proceedings of SPIE, 2001, 273~281.
[20]A. Elgammal, R. Duraiswami, D. Harwood, etc.Background and foreground modeling using nonparametric kernel density estimation for visual surveillance [C].Proceedings of the IEEE, 2002, 138~146.
[21]C. Wren, A. Azarbayejani, T. Darrell, etc.Pfmder:Real-time Tracking of the Human Body [C].IEEE Transactions on Pattern Analysis and Machine Intelligence, 1997, 19(7), 780~785.
[22]C. Stauffer, E. Grimson.Adaptive background mixture models for real-time tracking [C], in Proc. IEEE Conf. on Computer Vision and Pattern Recognition, Fort Collins, CO, vol. 02, 1999, 246~252.
[23]P. Kaew, Tra Pong,R. Bowden.An Improved Adaptive Background Mixture Model for Real-time Tracking with Shadow Detection [C].Proceedings of Second European Workshop on Advanced Video Based Surveillance Systems, 2001, 149~153.
[24]Gelman Andrew, Carlin John, S. Stern, and etc.Bayesian Data Analysis 2nd Edition [M], America, chapman & halu , 2003.
[25]Oncel Tuzel, Fatih Porikli, Peter Meer.A Bayesian Approach to Background Modeling [C].CVPR'05, 03, 58 ~63.
[26]J.Q. Shi, R. Murray-Smith, D.M. Titterington.Bayesian Regression and Classification Using Mixtures of Gaussian Processes [J].Adapt. Control Signal Process, 2003, 17, 1~16.
[27]Baldi Pierre, Long. D. Anthony.A Bayesian framework for the analysis of microarray expression data: regularized t-test and statistical inferences of gene changes [J].Bioinformatics, 17, 509~519.
[28]苏建.基于EM算法的基因芯片图像信号检测与分析[D],苏州大学硕士学位论文,2009.
[29]温津伟,罗四维,赵嘉利等.基于Bayesian的期望最大化方法BEM算法[J].计算机研究与发展,2001,38(7),821~824.
[30]俞奎,王浩,吴信东,等.贝叶斯网络的并行EM学习算法[J].模式识别与人工智能,2008,21(5),670~676.
[31]K. Smith, P. Quelhas, D. Gatica-Perez.Detecting abandoned luggage items in a public space [C].in PETS, 2006, 75~82.
[32]C. Sacchi, C. Regazzoni.A distributed surveillance system for detection of abandoned objects in unmanned railway environments [C].IEEE Transactions on Vehicular Technology, 2000, 49.
[33]P.T. Krahnstoever, T. Sebastian, A. Perera, etc.Multi-view detection and tracking of travelers and luggage in mass transit environments [C].VS2006, 2006, 67~74.
[34]Tian Yingli, Feris Rogerio, Hampapur Arun.Real-Time Detection of Abandoned and Removed Objects in Complex Environments [C].VS2008, Marseille, 2008.
[35]高大山.智能交通系统中运动目标检测方法的研究[D],清华大学硕士学位论文,2003.