摘要
随着科学技术的进步和计算机的广泛使用,数字图像处理技术已经渗透到人类生活的各个方面,并发挥着越来越重要的作用。图像分割就是指把图像分解成各具特性的区域并提取出感兴趣目标的技术和过程,它是由图像处理到图像分析的一个关键步骤,在计算机视觉、模式识别和医学图像处理等实际中得到了广泛的应用。
如何对彩色图像中的目标进行有效的分割是计算机视觉和图像分析的重点和难点,目前对彩色图像的分割方法主要分类为基于阈值的分割技术、基于边界的分割技术、基于区域特性的分割技术和结合特定理论工具的分割技术四类。
本文在传统的基于聚类算法的基础上,引入图像分层的理念,提出了一种基于二叉树结构的彩色图像分割方法,分层目的是将图像在不同分辨率下由粗到细地表达出来,基于分层的图像处理方法能够充分组合利用图像的全局与局部信息、空间与灰度信息.首先,对待分割图像采用最优阈值化方法获取R,G,B三个颜色空间的最佳阈值,从而得到三个颜色空间的二值图像,然后通过构造自适用二叉树进行一次粗分割提取目标区域,实现对图像的粗分割,自适应二叉树图像分块的基本思想是先把阈值化后的二值图像作为二叉树的根节点,以二值图像像素一致为基础,采用区域距离度量的方法进行区域分裂,在计算区域的距离时,我们需要考虑两个方面;区域的颜色距离和边缘距离,从而构造图像的分裂二叉树,得到图像的颜色一致区域。通过构造二叉树进行粗分割可以把图像上颜色一致的区域初步的分割开,而并不理解其中的含义。而且这样分割的图像往往呈方形,分割的结果不满足人们的视觉感知。因此本文将在此基础上采用C-均值聚类算法实现对图像粗分割后形成的叶子区域进行聚类。从而使分割后的图像更具有现实意义。本文使用Visual C++ 6.0实现了算法,实验结果证明,本文所提出的方法与传统的区域生长方法和C-均值聚类算法的相比,可以更好的实现目标图像和背景的分离。
With the development of the science and technology and widely usage of computer, technology of digital image processing have come into everyday life and have an important effect on everybody. Image segmentation is just to segment an image into different sub-images with different characters and get some interested objects. It is a key step from image process to image analysis, plays an important role in image engineering, and is applied in a lot of fields such as computer vision, pattern recognition, medical image and so on..
How to segment the target of the color image is the key and difficult of compute vision and Image Analysis. In recent years,People Separate image segmentation algorithms into 4 species.they are segmentation algorithms based on threshold , algorithms based on edge detection, algorithms based on regional characteristics,and algorithms combined with specific theoty.
On the basis of the the traditional clustering algorithm and the concept of tiered.an effective segmentation method based on binary tree is proposed in this paper. The aim is to express image from fine to coarse according to different resolution. Image processing based on the images layered approach can take full advantage of global and local information of the image, space and gray information. In the way of the the method used segmentation method based on binary tree, first, this method uses the optimal threshold to get the best threshold in the R.G..B color space.thus gained the binary image of R,G,B color space.Then a roughly extract of the color image is gotten by constructing the self-adapting binary tree. the basic idea of the segmentation method based on binary is we use the binary image as the Root node first, based on the consistency of the pixel. we split the picture by measure the distance of regions,there are two respect we need to consider: the color distance and the edge distance.so we can obtain the binary tree of the image and the ,construction of the images so as to split the binary tree and the color coherent region. now the image is segmented to color coherent region segment but without understanding the meaning of the image.and the image was often square which dose not meet people's visual perception. After extracting, C_means clustering algorithm is used to improve the accuracy of the segmentation of the binary tree’s leaves. Experiments show that this method can separation of goals and background better compaare to the algorithms based on regional growth and clustering.
引文
[1]Y Rui,S H Thomas,S F Chang.Image retrieval:Past,present and future.Journal of Visual Communication and Image Representation,1999,10( 1) :39-62
[2]C C Hsu,W W Chu,R K Taira.A knowledge-based approach for retrieving images by content.IEEE Trans on Knowledge and Data Engineering,1996,8( 4) :522- 532
[3]D A Forsyth,JMalik,M M Fleck et al.Finding pictures of objects in large collections of images.In:Proc of the Int' l Workshop on Object Representation in Computer Vision II,ECCV-9 6.Cambridge:IS & T and SPIE,1996.335- 360
[4]WaterBalloons: A hybrid watershed Balloon Snake segmentation.Image and Vision Computing, Volume 26, Issue 7, 2 July 2008, Pages 905-912
[5]王元全,贾云得.一种新的心脏核磁共振图像分割方法,[J]计算机学报,2007,30(1):129-136
[6]Using anatomical knowledge expressed as fuzzy constraints to segment the heart in CT images Pattern Recognition, Volume 41, Issue 8, August 2008, Pages 2525-2540
[7]陶文兵,金海,一种新的基于图谱理论的图像阈值分割方法.[j]计算机学报, 2007,30(1):110-119
[8]JItten.Art of Color ( Kunst der Farbe) .Ravensburg,Germany:Otto Maier Verlag,1961.
[9] Y Wu,Y T Zhuang,Y H Pan.Image retrieval system for Web: Webscope-CBIR.In: IEEE Proc of the 11th Int'l Workshop on Database and Expert Systems Applications.Greenwich,London,UK: IEEE CS Press 2000.620-624
[10]SSclaroff,M L Cascia,L Taycheretal.Unifying textual and visual cues for content-based image retrieval on the World Wide Web.Computer Vision and Image Understanding,1999,75( 1) :86-98
[11]梁光明,孙即祥,李仲陶,唐玉鹏.基于区域主色特征的目标匹配算法[J].光电工程, 2008,(01):22-25
[12] Scholkopf B, Burges C J C, Smola A J, Advances in Kernel Metoods——Support Vector Learning.[M] Cambridge, MA: MIT Press,1999
[13] V Vapnik The Nature of Statistical Learning Theory[M].New York:Springer-Verlag,1995
[14]周献中.基于HSV颜色空间加权不变矩的台标识别[J].南京理工大学学报, 2005,2(29) : 363-367.
[15] Bottou L, Cortes C, Denker J, et al Comparison of Classifier Methods: A Case Study in Handw ritten Digit Recognition, A . Proc of the Int Conf on Pattern Recognition [C]. Jerusalem , 1994: 77-87
[16] Platt J C, Cristianini N, Shawe-Taylor J, Large margin DAGs for multiclass classification [C].Advances in Neural Information Processing Systems .2000.547-553
[17]田沛,范瑾,李亮,张兆祥基于类间方差和形态学的一类生物特征识别.[J]清华大学学报(自然科学版).2007:47(S2):1747-1750.
[18] 1 LIN Kaiyan,XU Li hong,WU Junhui.A Survey on Color Image Segmentation Techniques [J].Journal of Image and Graphics,2005:10(1):1-11
[19]WANG Yuelan,ZENG Yingsheng. The Study on the Information Fusion Used for Color Images Segmentation [J]. CHINESE JOURNAL OF COMPUTERS, 2000,23(7):763-767.
[20]Kurugollu F,Sankur B. Harmanci A E.Color Image Segmentation Using Histogram Multithresholding and Fusion.Image and Vision Computing 2001.19(13):915-928.
[21] H D Cheng,J Li. Fuzzy homogeneity and scale - space approach to colorimage segmentation[J]. Pattern Recognition 2003(36):1545-1562.
[22] H D Cheng, ,X H Jiang,JingliWang. Color image segmentation based homogram thresholding and regionmerging[J]. Pattern Recognition 2002(35):373-393.
[23]杨晖,郑刚,张仁杰,张军勇.基于遗传算法的指纹图像分割算法.[J]光学技术.2007,33(5):772-774
[24]YE Qixiang,GAO Wen,WANG Weiqiang,HUANG Tiejun. A Color Image Segmentation Algorithm by Using Color and Spatial Information [J]. Journal of Software,2004,15(4):522 - 530.
[25]刘直芳,王运琼,游志胜.简单有效的运动汽车投影阴影分割算法[J].光学学报2007.27(5):835-840
[26]李旭超,朱善安,朱胜利,基于小波域层次Markov模型的图像分割[J]中国图象图形学报2007,12(2):308-314.
[27]张志勇,施智平,张素兰,史忠植.基于小波变换和ICA的运动目标分割[J]计算机辅助设计与图形学学报.2007.19(4):448-453
[28] D. Karatzas. A. Antonacopoulos.Colour text segmentation in web images based on human perception.Image and Vision Computing, Volume 25, Issue 5, 1 May 2006, Pages 564-577
[29] Zoltan Kato .Segmentation of color images via reversible jump MCMC sampling Image and Vision Computing, In Press, Corrected Proof, Available online 19 December 2006,
[30]付峰,应义斌.生物图像阈值分割方法的研究[J].浙江大学学报,2003,29(1):108-112. [31 Milind M. Mushrif, Ajoy K. Ray Color image segmentation: Rough-set theoretic approachPattern Recognition Letters, Volume 29, Issue 4, 1 March 2008, Pages 483-493
[32]A Garcia. Probability and Random Process for Electrical Engineering[M]. Reading,MA:AddisonWesley,1989.
[33] C Huang, T Cheng, C Chen. Color Image Segmentation Using Scale Space Filter andMarkov Random Field[J]. pattern Recognition,1992. 25(10):1217-1229.
[34]Nalwa V S,Binford T O.On Detecting Edges.IEEE Trans on Pattern Analysis and Machine Intelligence,1986,8(6):699-714
[35]Haralick R M.Digital Step Edges from Zero Crossing of Second Directional Derivatives.IEEE Trans on Pattern Analysis and Machine Intelligence,1984,6(1):58-68
[36]Matalas L,Benjamin R,Kitney R.An Edge Detection Technique Using the Facet Model and Parameterized Relaxation Labeling.IEEE Trans on Pattern Analysis and Machine Intelligence,1997,19(4):328-341
[37]Canny J.A Cmputational Approach to Edge Detection.IEEE Trans Pattern Analysis and Machine Intelligence,1986,8(6):679-698
[38]Wikinson M H F.Optimizing Edge Detectors for Robust Automatic Threshold Selection:Coping with Edge Curvature and Noise.Graphic Models and Image Processing,1998,60(4):385-401
[39]Witkin A.Scale-space filtering:A new approach to multi-scale description.In:Proceedings of IEEE International Conference on Acoustics,Speech,and Signal Processing.1984.9:150-153
[40]Chanda B,Kundu M K K,Padmaha V Y.A Multi-scale Morphologic Edge Detector.Pattern Recognition,1998,31(10):1469-1478
[41]Wang Y P,Lee S L.Scale-Space Derived from B-Splines.IEEE Trans on Pattern Analysis and Machine Intelligence,1998,20(10):1040-1055
[42]Elder J H,Zucher S W.Local Scale Control for Edge Detection and Blur Estimation.IEEE Trans on Pattern Analysis and Machine Intelligence,1998,20(7):699-716
[43]周培德,付梦印,黄源水,刘羿彤.红外图像边缘提取的算法[J].兵工学报2007:28(5).524-527
[44]Ahuji N.A Transform for Multiscale Image Segmentation by Integrated Edge and Region Detection.IEEE Trans on Pattern Analysis and Machine Intelligence,1996,18(12):1211-1234
[45]Tabb M,Ahuja N.Multiscale Image Segmentation by Integrated Edge and Region Detection.IEEE Trans on Image Processing,1997,6(5):642-655
[46]Montanvert A,Meer P,Rosenfeld A.Hierarchical Image Analysis Using Irregular Tessellations.IEEE Trans on Pattern Analysis and Machine Intelligence,1989,11(4):307-316
[47]Koster A A E,Vincken K L,De Graaf C N et al.Heuristic Linking Models in Multi-scale Image Segmentation.Computer Vision and Image Understanding,1997,65(3):382-402
[48]Staib L H,Duncan J S.Boundary Finding with Parametrically Deformable Models.IEEE Trans on Pattern Analysis and Machine Intelligence,1992,14(11):1061-1075
[49]Gashtasby A.Design and Recovery of 2-D and 3-D Shapes Using Retional Gaussian Curves and Surfaces.International Journal of Computer Vision,1993,10(3):233-256
[50]Legault R,Suen C Y.Optimal Local Weighted Averaging Methods in Contour Smoothing.IEEE Trans on Pattern Analysis and Machine Intelligence,1997,19(9):801-817
[51] Byoung-Ki Jeon , Yun-Beom Jung , Ki-Sang Hong.Image segmentation by unsupervised sparse clustering. Pattern Recognition Letters 27 (2006) 1650–1664
[52] TU chiyama, M.Arbib,Color Image Segmentation Using Competitive Learning[J]. IEEE Trans. On Pattern analysis and Machine Intelligence, 1994,16(12):1197-1206.
[53] T Uchiyama ,M Arbib, Color I mage Segmentation Using Competitive Learning[ J ]. IEEE Trans . On Pattern analysis andMachine Intelligence1994, 16 (12) : 1197-1206.
[54] Y Lim, S Lee . On the Color I mage Segmentati on Algorithm Based on Threshold and the Fuzzy c - means Technique[ J ]. Pattern Recognition,1990, 23 (9) : 935-952.
[55] L lucchese, SMitra . Unsupervised Segmentation of Col or I mage Based on K - means Clustering in the Chromacity Plane[A ]. Proceeding of IEEE Workshop on Content - based Access of I mage and Video Libraries(CBA I VL′99) [ C ]. Fort Collins, CO, 1999. 74 - 78 . [56 ] O Verevka,Buchann,Local k - meansAlgorithm for Col or I mage Quantization[A ]. Proceeding of Graphics Interface[ C ]. 1995. 128-136. [57 ] T yamazaki . I ntroduction of EM Algorithm into Color I mage Segmentati on[A ]. Pr oceedings of 2nd IEEE I nternati onal Conference on I ntelligent Processing Systems( I CIPS′98) [ C ]. 1998. 368 - 371.
[58] N Kehtarnavas, J Monaco, J Nimscheck and AWeeks . Col or I mage Segmentati on UsingMulti - Scale Clustering[A ]. Proceeding of IEEE South-west Symposium on I mage Analysis and I nterpolati on[ C ]. Tucs on,AZ, 1998. 142 - 147.
[59]樊云,王润生.面向内容检索的彩色图象分割[ J ].计算机研究与发展, 2002, 39 (3) : 376 - 381.
[60]潘晨,顾峰.基于3D直方图的彩色图象分割方法[ J ].中国图像图形学报, 2002, 7A (8) : 800-805.
[61] H D Cheng, ,X H Jiang,JingliWang. Color image segmentation based homogram thresholding and regionmerging[J]. Pattern Recognition 2002(35):373-393.
[62] H D Cheng,J Li. Fuzzy homogeneity and scale - space approach to colorimage segmentation[J]. Pattern Recognition 2003(36):1545-1562.
[63] TD Pham,H Yan. color image segmentation using fuzzy integral andmountain clustering[J]. Fuzzy Sets and System 1999(107) :121-130.
[64]林开颜,徐立鸿,吴军辉.快速模糊C均值聚类彩色图像分割方法[J].中国图像图形学报,2004,9A (2):159-164.
[65]刘昌进,郭立,朱俊株,等.基于去降Mallat离散小波变换的彩色图象分割[J].计算机工程与应用,2003. 11:93-95.
[66]赵建伟,王朋,刘重庆.基于小波变换的分水岭图象分割方法[J].光子学报,2003. 5:601-604.
[67] Byung- Gyu Kim, Jae - Ick shim,Dong- Jo park. Fast image segmentation based onmulti- resolution analysis andwavelets[J]. Pattern recognition Letters2003(24) :2995-3006.
[68]H.Rom,S.Peleg,“Image representation using voronoi tessellation:adaptive and secure”,IEEE Comput.Vision Pattern Recognition(1988)282–285.
[69]J.Liu,Y.Yang,“Multi-resolution color image segmentation”,IEEE Trans.Pattern Anal. Mach.Intell.16(7)(1994)689–700.
[70]B.Sabata,F.Arman,J.Aggarwal,“Convergence of fuzzy-pyramidalgorithms”,J.Math.Imaging Vision 4(1994)291–302.
[71]R.Marfil,L.Molina-Tanco,A.Bandera,J.A.Rodríguez,andF.Sandoval“Pyramid Segmentation Algorithms Revisited”,Pattern Recognition,Vol.39,No.8,pp.1430-1451,2006.
[72]D.Prewer,L.J.Kitchen,“Soft image segmentation by weighted linkedpyramid”,Pattern Recognition Lett.22(2)(2001)123–132.
[73]F.Ziliani,B.Jensen,“Unsupervised Image Segmentation Using The Modified Pyramidal Linking Approach”,International Conference on Image Processing,vol.3,1998,pp.303–307.
[74]严蔚敏,吴伟民,数据结构,清华大学出版社,1997
[75]Douglas B.West.“Introduction to Graph Theory”,China Machine Press,2004.
[76]T.H.Cormen,C.E.Leiserson,R.L.Rivest,and C.Stein,“Introduction to Algorithm”,2nd Edition,The MIT Press,Higher Education Press,Beijing,China,2002.
[77]S.Julier, J.K.Uhlmann, A General Method for Approximating Nonlinear Transformation of Probability Distribution[C], Tech. rep., RRG, Dept, of Engineering Scence, University of Oxford, 1996.
[78]J.E.Handschin, Monto Carlo techniques for prediction and filtering of nonlinear stochastic processes [J], Automatic 1969,6:555-563
[79]J.R.Carpenter, P.Clifford, P.Fearnhead, Efficient implementation of particle filters for non-linear systems[C], 4th Interim Report, DRA contract WSS/U1172, Department of Statistics, Oxford University, 1997
[80]A.M. Sanjeev, M.Simon, G.Neil, T.Clapp, A Tutorial on Particle Filters for Online Nonlinear/Non一Gaussian Bayesian Tracking [J], IEEE Transaction on Signal Processing, 2002, 50(2):174一188
[81] Lin K Y, XuL H, Wu J H. A fast fuzzyC2means clustering for color image segmentation [J]. Journal of Image and Graphics, 2004, 9(2):159-163.