摘要
随着经济的发展,人民生活水平的提高,汽车数量的增长速度远远超过道路基础施建设的速度,城市公路交通系统的压力不断加大,智能交通系统作为一种新的交通管理技术,受到全世界范围的高度重视。本文主要研究智能交通系统中的运动车辆视频检测问题,在背景提取与更新、阴影抑制和车辆轮廓提取等方面提出了自己的算法。概括起来,本文的主要研究内容如下:
首先,我们通过改进现有算法,提出了一种基于块特征的背景提取与更新方法。该方法将监控区域分块,通过计算时域上相邻帧对应位置的块的相似程度来判断当前块是否处于变化状态,然后将被判为静止状态的块置入一个时域上的缓冲区,使用缓冲区内储存的块提取背景并实时进行更新。该算法区别于传统的按像素计算的背景模型算法,基于块的思想使得计算复杂程度大大降低,同时仍能取得较好的背景效果。
然后,我们提出了一种结合颜色和纹理信息的去阴影算法,除去阴影后得到车辆前景目标。该算法分为两步,第一步是使用颜色信息从前景目标中判断出阴影区域,进行去除,第二步是使用纹理信息获取车辆和阴影的边缘,再利用第一步的结果和阴影边缘本身的特性,将阴影的边缘去除。实验证明,该方法适用于任何方向的阴影,相比传统算法更有优势。但是,对于少数车辆,使用该方法仍然得不到完整的轮廓。
最后,本文又提出了一种基于先验形状约束的车辆检测的方法,以前面的去阴影算法得到的结果为初始值,使用先验形状信息把车辆轮廓修复完整。具体实现时,将先验的车辆形状信息作为约束融入主动轮廓模型,采用形状配准和水平集方法演化曲线,直到曲线收敛。实验证明,使用该算法能弥补前一种算法的缺陷,获得车辆的准确轮廓。
With the development of city and economy and the improvement of the people's living standard, the number of vehicles on the road increased quickly, and the burden of the road transportation system becomes higher and higher. Therefore, more attention has been paid to the vision-based intelligent transportation system. In this paper, we do some researches on vehicle detection, which is the key technology of the intelligent transportation system, and propose algorithms on background extraction and update, and shadow suppression. Research works in this paper are summarized as follows:
Firstly, we propose a background extraction and update method based on block features by improving an existing algorithm. The method divides images into blocks, and computes statistical likelihood for each block in the time domain. Blocks with slight change are classified as the static blocks and put into a buffer. These blocks are used to extract and update the background image. Compared with the background model based on pixel, this method can reduce the computing complex significantly, and can also get good result.
Secondly, we present an algorithm to do vehicle segmentation and cast shadow removal using the color and texture information. In the first step, the method uses color information to detect shadows in objects. In the second step, texture features are used to find edges of vehicles and shadows. Then we combine characters of the shadow edges and the result in the first stage to remove the shadow edges. Experimental results show that the method can cope with shadows in any directions. But for a few of vehicles, this method still can't obtain the whole contour.
Finally, a vehicle detection method based on the prior shape knowledge is proposed. This method uses the result of the shadow removal algorithm as the initial contour, and restores the vehicle contour with prior shape knowledge. An implicit shape model is built and an active contour energy function with the restriction of the existing shape priors is constructed. Then we apply the shape alignment and level set method to evolving the initial contour until convergence. Results in the experiments show that we can overcome the defect of the shadow removal algorithm using this method and get the precise contours of vehicles.
引文
[1]M.Tomizuka.Automated highway systems-an intelligent transportation system for the next century.The 1997 1 st IEEE/ASME International Conference on Advanced Intelligent Mechatronics,AIM'97,1997:1997.
[2]陈旭梅,于雷,郭继孚.美国智能交通系统ITS的近期发展综述.中外公路,2003,23[2]:9-12.
[3]B.K.P.Horn,B.G.Schunck.Determining Optical Flow.Artificial Intelligence,1981,17[1-3]:185-203.
[4]G.Adiv.Determining three-dimensional motion and structure from optical flow generated by several moving objects.IEEE Transactions on Pattern Analysis and Machine Intelligence,1985,7:384-401.
[5]J.L.Barron,D.J.Fleet,S.S.Beauchemin.Performance of optical flow techniques.International Journal of Computer Vision,1994,12[1]:43-77.
[6]B.Gloyer,H.K.Aghajan,K.Y.Siu,et al.Video-based freeway-monitoring system using recursive vehicle tracking.Proceedings of SPIE,1995,2421:173.
[7]W.Long,Y.H.Yang.Stationary background generation:An alternative to the difference of two images.Pattern Recognition,1990,23[12]:1351-1359.
[8]N.Friedman,S.Russell.Image segmentation in video sequences:A probabilistic approach.Proc.ofthe 13th Conf.on Uncertainty in Artificial Intelligence,1997:175-181.
[9]C.Stauffer,W.E.L.Grimson.Adaptive background mixture models for real-time tracking.Proceedings of the IEEE Computer Society Conference on Computer Vision and Pattern Recognition,1999,2:246-252.
[10]P.KaewTraKulPong,R.Bowden.An Adaptive Visual System for Tracking Low Resolution Colour Targets.British Machine Vision Conf,2001:243-252.
[11]A.Elgammal,D.Harwood,L.Davis.Non-parametric model for background subtraction.FRAME-RATE Workshop,IEEE,1999.
[12]D.R.Magee.Tracking multiple vehicles using foreground,background and motion models.Image and Vision Computing,2004,22[2]:143-155.
[13]R.Tan,H.Huo,J.Qian,et al.Traffic Video Segmentation Using Adaptive-K Gaussian Mixture Model.Advances In Machine Vision,Image Processing,and Pattern Analysis Lecture Notes In Computer Science,2006,4153:125-134.
[14]L.He,Z.Zhang.Real-time whiteboard capture and processing using a video camera for teleconferencing.Acoustics,Speech,and Signal Processing,2005.Proceedings.(ICASSP'05).IEEE International Conference on,2005,2.
[15] N. Otsu. A threshold selection method from gray-level histograms. IEEE Transactions on Systems, Man, and Cybernetics, 1979,9[1]: 62-66.
[16] A. Prati, R. Cucchiara, I. Mikic, et al. Analysis and Detection of Shadows in Video Streams: A Comparative Evaluation. Proc. Third Workshop Empirical Evaluation Methods in Computer Vision. Computer Vision and Pattern Recognition, 2001.
[17] I. Mikic, P. C. Cosman, G. T. Kogut, et al. Moving shadow and object detection in traffic scenes. Pattern Recognition, 2000. Proceedings. 15th International Conference on, 2000,1.
[18] T. Horprasert, D. Harwood, L. S. Davis. A statistical approach for real-time robust background subtraction and shadow detection. Proc. IEEE ICCV, 99:1-19.
[19] I. Haritaoglu, D. Harwood, L. S. Davis. W4: Real-Time Surveillance of People and Their Activities. 2000.
[20] R. Cucchiara, C. Grana, M. Piccardi, et al. Detecting moving objects, ghosts, and shadows in video streams. Pattern Analysis and Machine Intelligence, IEEE Transactions on, 2003,25[10]: 1337-1342.
[21] J. Stander, R. Mech, J. Ostermann. Detection of moving cast shadows for object segmentation. Multimedia, IEEE Transactions on, 1999,1[1]: 65-76.
[22] J. W. Hsieh, W. F. Hu, C. J. Chang, et al. Shadow elimination for effective moving object detection by Gaussian shadow modeling. Image and Vision Computing, 2003,21[6]: 505-516.
[23] A. Yoneyama, C. H. Yeh, C. C. J. Kuo. Moving cast shadow elimination for robust vehicle extraction based on 2D joint vehicle/shadow models. Proceedings. IEEE Conference on Advanced Video and Signal Based Surveillance, 2003., 2003:229-236.
[24]袁基炜,史忠科．一种运动车辆的阴影消除新方法．西安交通大学学报,2005,39[6]．
[25] B. V. Funt, M. S. Drew, M. Brockington. Recovering shading from color images. ECCV-92: Second European Conference on Computer Vision: 124-132.
[26] T. Gevers, A. W. M. Smeulders. Color-based object recognition. Pattern Recognition, 1999, 32[3]: 453-464.
[27] E. Salvador, A. Cavallaro, T. Ebrahimi. Shadow identification and classification using invariant colormodels. Acoustics, Speech, and Signal Processing, 2001. Proceedings.(ICASSP'01). 2001 IEEE International Conference on, 2001,3.
[28] N. Martel-Brisson, A. Zaccarin. Moving cast shadow detection from a gaussian mixture shadow model. Proceedings of IEEE Conference on Computer Vision and Pattern Recognition, 2005: 643-648.
[29] D. Xu, X. Li, Z. Liu, et al. Cast shadow detection in video segmentation. Pattern Recognition Letters, 2005, 26[1]: 91-99.
[30]G.S.K.Fung,N.H.C.Yung,G.K.H.Pang,et al.Effective moving cast shadow detection for monocular color imagesequences.Image Analysis and Processing,2001.Proceedings.11th International Conference on,2001:404-409.
[31]W.Zhang,X.Z.Fang,X.Yang.Moving cast shadows detection based on ratio edge.Proceedings of the 18th International Conference on Pattern Recognition(ICPR'06)-Volume 04,2006:73-76.
[32]张伟,基于视觉的运动车辆检测与跟踪,博士学位论文,上海交通大学,上海,2007.
[33]M.Kass,A.Witkin,D.Terzopoulos.Snakes:Active contour models.International Journal of Computer Vision,1988,1[4]:321-331.
[34]L.D.Cohen.On active contour models and balloons.CVGIP:Image Understanding,1991,53[2]:211-218.
[35]C.Xu,J.L.Prince.Snakes,shapes,and gradient vector flow.Image Processing,IEEE Transactions on,1998,7[3]:359-369.
[36]B.Das,S.Banerjee.Homogeneity induced inertial snake with application to medical image segmentation.Computer-Based Medical Systems,2004.CBMS 2004.Proceedings.17th IEEE Symposium on,2004:304-309.
[37]S.R.Gunn,M.S.Nixon.A robust snake implementation;a dual active contour.IEEE Transactions on Pattern Analysis and Machine Intelligence,1997,19[1]:63-68.
[38]T.McInerney,D.Terzopoulos.T-snakes:Topology adaptive snakes.Medical Image Analysis,2000,4[2]:73-91.
[39]R.Ronfard.Region-based strategies for active contour models.International Journal of Computer Vision,1994,13[2]:229-251.
[40]S.Menet,P.Saint-Marc,G.Medioni.B-snakes:Implementation and application to stereo.Proceedings DARPA,1990,720726.
[41]S.Osher,J.Sethian.Fronts propagating with curvature dependent speed:algorithms based on the Hamilton-Jacobi formulation.Journal of Computational Physics,1988,79[1]:12-49.
[42]V.Caselles,R.Kimmel,G.Sapiro.Geodesic Active Contours.International Journal of Computer Vision,1997,22[1]:61-79.
[43]岑峰,戚飞虎,曾文珺.基于边缘吸引力场正则化的短程线主动轮廓模型.电子学报,2003,31[1]:17-20.
[44]A.Yezzi Jr,S.Kichenassamy,A.Kumar,et al.A geometric snake model for segmentation of medical imagery.Medical Imaging,IEEE Transactions on,1997,16[2]:199-209.
[45]李海云,李光颖,王筝.一种基于水平集的脊柱MRI图像分割算法的研究.北京生物医学工程,2004,23[3]:178-180.
[46]朱付平,田捷,林瑶等.基于Level Set方法的医学图像分割.软件学报,2002,13[9]:1866-1872.
[47]R.Malladi,J.A.Sethian.A unified approach to noise removal,image enhancement,and shaperecovery.Image Processing,IEEE Transactions on,1996,5[11]:1554-1568.
[48]李俊,杨新.基于Mumford-Shah模型的快速水平集图像分割方法.计算机学报,2002,25[11]:1175-1183.
[49]陆东莹,庄天戈.基于Level Set方法的低对比度医学图像分割.上海交通大学学报,2006,40[8]:1444-1447.
[50]M.Droske,B.Meyer,M.Rumpf,et al.An adaptive level set method for medical image segmentation.Proc.of the Annual Symposium on Information Processing in Medical Imaging,2001.
[51]S.Osher,M.Burger,D.Goldfarb,et al.Using geometry and iterated refinement for inverse problems(1):Total variation based image restoration.Department of Mathematics,UCLA,LA,CA,2004:04-13.
[52]于慧敏,徐艺,刘继忠等.基于水平集的多运动目标时空分割与跟踪.中国图象图形学报,2007,12[7]:1218-1223.
[53]Y.H.-m.XU Yi.Contour-Based Motion Segmentation Using Few Priors.International Conference on Signal Processing,2006:1376-1379.
[54]于慧敏,尤育赛.基于水平集的多运动目标检测和分割.浙江大学学报:工学版,2007,41[3]:412-417.
[55]徐艺,基于PDEs的运动目标检测与轮廓跟踪算法研究,浙江大学,杭州,2006.
[56]尤育赛,关于数字图像处理在运动目标检测和医学检验中若干应用的研究,浙江大学,杭州,2005.
[57]http://mountains.ece.umn.edu/～guille/matlab/cursohtml/Numerical%20algorithms.htm.
[58]S.Osher,R.Fedkiw,Level Set Methods and Dynamic Implicit Surfaces,NY:Springer Press,2002.
[59]D.Adalsteinsson,J.Sethian.A fast level set method for propagating surfaces.Journal of Computational Physics,1995,118[2]:269-277.
[60]J.A.Sethian,Level set methods and fast marching methods,Cambridge University Press Cambridge,1999.
[61]R.T.Whitaker.A Level-Set Approach to 3D Reconstruction from Range Data.International Journal of Computer Vision,1998,29[3]:203-231.
[62]解可新,韩立兴,林友联,最优化方法,天津大学出版社,1997.
[63]P.Wolfe.Methods of nonlinear programming.Recent Advances in Mathematical Programming,1963:67-86.
[64]J.Abadie,J.Carpentier.Generalization of the Wolfe Reduced Gradient Method to the case of Nonlinear Constraints.New York,1969:37-47.
[65]段先华,夏德深.基于椭圆约束分割心脏MRI图像的水平集模型.计算机工程,2007,33[16]:14-16.
[66]T.F.Chan,L.Vese.A level set algorithm for minimizing the Mumford-Shah functional in image processing.IEEE/Computer Society Proceedings of the 1st IEEE Workshop on "Variational and Level Set Methods in Computer Vision,2001:161-168.
[67]A.L.Yuille,P.W.Hallinan,D.S.Cohen.Feature extraction from faces using deformable templates.International Journal of Computer Vision,1992,8[2]:99-111.
[68]M.P.Dubuisson,S.Lakshmanan,A.K.Jain.Vehicle segmentation using deformable templates.IEEE Trans.Pattern Anal.and Machine Intelligence,1996,18[3]:293-308.
[69]T.F.Cootes,C.J.Taylor,D.H.Cooper,et al.Active shape models-their training and application.Computer Vision and Image Understanding,1995,61[1]:38-59.
[70]M.E.Leventon,W.E.L.Grimson,O.Faugeras.Statistical shape influence in geodesic active contours.Computer Vision and Pattern Recognition,2000.Proceedings.IEEE Conference on,2000,1.
[71]A.Litvin,W.C.Karl.Using shape distributions as priors in a curve evolution framework.Acoustics,Speech,and Signal Processing,2004.Proceedings.(ICASSP'04).IEEE International Conference on,2004,3.
[72]M.Rousson,N.Paragios.Shape priors for level set representations.Proc.of the Europ.Conf.on Comp.Vis,2002,2351:78-92.
[73]T.F.Cootes,A.Hill,C.J.Taylor,et al.The use of active shape models for locating structures in medical images.Image and Vision Computing,1994,12[6]:355-366.
[74]M.Rousson,N.Paragios.Prior Knowledge,Level Set Representations & Visual Grouping.International Journal of Computer Vision:1-13.