面向医学应用的纹理图像分割方法研究
|
摘要
基于内容的图像检索技术通过分析图像的颜色、纹理、形状等视觉特征，从图像库中查找含有特定特征的图像。它克服了传统的基于文本的图像检索方法的不足，融合了图像处理、图像识别和图像数据库等领域的技术成果，从而可以提供更有效的检索手段。对于基于内容的图像检索来说，图像分割是最重要的图像预处理技术，只有经过特征提取和相应的图像分割之后，图像才能入库，进而实现基于不同“内容”的图像检索。
在对图像的研究和应用中，人们往往仅对图像中的某些部分(即目标)感兴趣。这些目标一般对应图像中特定的、具有独特性质的区域。为了识别和分析图像中的目标，需要将它们从图像中提取出来，在此基础上才有可能进一步对图像进行分析利用，以及对目标进行测量。图像分割指的就是把图像分成各具特性的区域并提取出感兴趣目标的技术和过程。它是从图像处理到图像分析的关键步骤，也是进一步图像理解的基础。
本文首先对常用的图像分割方法进行了详细的分类介绍和综述，并在此基础上，阐述了图像分割技术在生物医学图像研究中的应用和研究重点。接着分类研究了纹理图像的分割方法，并针对医学图像，特别是人体内脏的医学图像所包含的纹理信息的特点，利用共生矩阵及其各种纹理特征参量对一幅人体肝脏的CT图像进行了纹理分析。分析试验的结果表明，上述参量能够很好的反映人体内脏图像中的纹理变化。基于这一分析结果，本文利用灰度共生矩阵及其特征参量，提取图像的纹理特征，并利用所提取的纹理特征设计一种结合灰度和纹理特征的区域增长算法以实现纹理图像的分割。最后利用程序设计语言对所设计的算法进行编程实现。
对图像分割实验的结果进行测量比较，结果显示该算法对特定类型的医学图像具有较好的分割效果。该算法的设计实现及图像分割的结果表明，多种信息的结合是解决图像分割问题的一个有效的新思路。
CBIR (Content Based Image Retrieval) is characterized by the ability of the system to retrieve relevant images based on their visual contents, such as color, texture, shape, rather than by using atomic attributes or keywords assigned to them. Image segmentation is the most important preprocessing technology of CBIR, only through feature extraction and image segmentation can CBIR be implemented successfully.
We are always interested in some parts (targets) of a image when we research on it. These targets are corresponding to some special regions in the image. These targets must be extracted from image before we could recognize and analysis them. Image segmentation is the technology and process that divide image into some regions, which have their own characteristics. Image segmentation is the key step from image processing to image analysis, and it is also the foundation of image understanding.
In this paper, we make a summary of image segmentation methods firstly, and then recount its applications and research in biomedicine. Secondly, we review the segmentation methods of image integrated with texture information, and then we make a texture analysis on a human liver CT image through texture feature parameter based on co-occurrence matrix. The experiment result shows that these parameters can reflect texture character of different region in human liver CT image. Based on the result, a region-growing algorithm is proposed, through extracting texture parameter that based on co-occurrence matrix from image. This method makes use of gray and texture information. Finally, the method is implemented by program language.
The result of experimentation shows the method that is designed in the paper is effective in segmentation of some medical images, and the idea, which integrates many kinds of information into segmentation method, will be an important direction in the research on image segmentation.
在对图像的研究和应用中，人们往往仅对图像中的某些部分(即目标)感兴趣。这些目标一般对应图像中特定的、具有独特性质的区域。为了识别和分析图像中的目标，需要将它们从图像中提取出来，在此基础上才有可能进一步对图像进行分析利用，以及对目标进行测量。图像分割指的就是把图像分成各具特性的区域并提取出感兴趣目标的技术和过程。它是从图像处理到图像分析的关键步骤，也是进一步图像理解的基础。
本文首先对常用的图像分割方法进行了详细的分类介绍和综述，并在此基础上，阐述了图像分割技术在生物医学图像研究中的应用和研究重点。接着分类研究了纹理图像的分割方法，并针对医学图像，特别是人体内脏的医学图像所包含的纹理信息的特点，利用共生矩阵及其各种纹理特征参量对一幅人体肝脏的CT图像进行了纹理分析。分析试验的结果表明，上述参量能够很好的反映人体内脏图像中的纹理变化。基于这一分析结果，本文利用灰度共生矩阵及其特征参量，提取图像的纹理特征，并利用所提取的纹理特征设计一种结合灰度和纹理特征的区域增长算法以实现纹理图像的分割。最后利用程序设计语言对所设计的算法进行编程实现。
对图像分割实验的结果进行测量比较，结果显示该算法对特定类型的医学图像具有较好的分割效果。该算法的设计实现及图像分割的结果表明，多种信息的结合是解决图像分割问题的一个有效的新思路。
CBIR (Content Based Image Retrieval) is characterized by the ability of the system to retrieve relevant images based on their visual contents, such as color, texture, shape, rather than by using atomic attributes or keywords assigned to them. Image segmentation is the most important preprocessing technology of CBIR, only through feature extraction and image segmentation can CBIR be implemented successfully.
We are always interested in some parts (targets) of a image when we research on it. These targets are corresponding to some special regions in the image. These targets must be extracted from image before we could recognize and analysis them. Image segmentation is the technology and process that divide image into some regions, which have their own characteristics. Image segmentation is the key step from image processing to image analysis, and it is also the foundation of image understanding.
In this paper, we make a summary of image segmentation methods firstly, and then recount its applications and research in biomedicine. Secondly, we review the segmentation methods of image integrated with texture information, and then we make a texture analysis on a human liver CT image through texture feature parameter based on co-occurrence matrix. The experiment result shows that these parameters can reflect texture character of different region in human liver CT image. Based on the result, a region-growing algorithm is proposed, through extracting texture parameter that based on co-occurrence matrix from image. This method makes use of gray and texture information. Finally, the method is implemented by program language.
The result of experimentation shows the method that is designed in the paper is effective in segmentation of some medical images, and the idea, which integrates many kinds of information into segmentation method, will be an important direction in the research on image segmentation.
引文
[1] Gudivada, V.N. Raghavan, V.V. Content based image retrieval systems. Computer, 1995, 28 (9): 18～22
[2] Swain M.J. Ballard D.H. Indexing via color histograms. In proceedings of 3rd Int. Conf. Computer Vision, 1990:390～393
[3] 李向阳，鲁东明，潘云鹤．基于颜色的图像数据库检索方法的研究．计算机研究与发展，1999，3(6)：359～363
[4] Smith J.R. Integrated Spatial and Feature Image System: Retrieval, Analysis and Compression: [D]. New York: Columbia University, 1997
[5] Zhang Lei, Lin Fuzong, Zhang Bo. A CBIR Method Based on Color-Spatial Feature TENCON 99. In Proceedings of the 10th IEEE Region Conference, 1999, 1: 166～169
[6] Haralick R M, Shanmugam K D. Textural Features for Image Classification. IEEE Trans on SMC, 1973, 3: 610～621
[7] Carter P.H. Texture Discrimination Using Wavelets. In proceedings of the 1993 IEEE Computer Society Conference in Computer Vision and Pattern Recognition, 1993:640～641
[8] Ballard D.H. Generalizing the Hough Transform to Detect Arbitrary Shapes. Pattern Recognition, 1981, 13(2): 111～122
[9] Kass M, Witkin A, Terzopoulous D. Snakes: active contour models. In Proceedings of the 1 st International Conference on Computer Vision. London: IEEE Computer Society Press, 1987:259～268
[10] 柳伟，李国辉，曹莉华．一种基于内容的图象检索方法的实现．中国图象图形学报，1998，3(4)：304～308
[11] Amoid W M, Marce W, Simone S et al. Content-based image retrieval at the end of the early years. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2000, 22(12): 1349～1379.
[12] Hosoba M, et al. PACS Workstation for Computer Assisted Image Diagnosis. Processing SPIE Medical Imaging Conference Ⅳ, 1990
[13] 章毓晋．图像分割．北京：科学出版社，2001：2～4
[14] 章毓晋．图像分割．北京：科学出版社，2001：9～14
[15] Haralick R M. Digital step edges from zero crossing of second directional derivative. IEEE-PAMI, 1984, 6(1): 58～66
[16] Ostu N. A threshold selection method from gray-level histogram. IEEE Trans on SMC, 1979, 9:62～66
[17] O'gorman L, Sanderson A C. The converging square algorithm: an efficient method for locating peaks in multidimensions. IEEE-PAMI, 1984, 6:280～288
[18] O'gorman L, Sanderson A C. The wedge filter technique for convex boundary estimation. IEEE-PAMI, 1985, 7:326～332
[19] Hong T H. Rosenfeld A. Compact region extraction using weighted pixel linking in a pyramid. 1984, 6(2): 222～237
[20] Ron Schoenmakers. Integrated Methodology for Segmentation of Large Optical Satellite Images. In Land
Application of Remote Sensing. Luxembourg, Italy, 1995
[21] Kass M, Witkin A, Terzopoulos D. Snakes: Active Contour Models. Int. J. Computer Vision, 1988, 1(4): 321～331
[22] Chiou G I, Hwang J N. A Neural Network-based Stockhastic Active contour model (NNS-SNAKE) for contour finding of distinct features. IEEE Trans. On Medical Imaging, 1995, 4:1407～1416
[23] Gunn S R, Nixion M S. A Robust Snake Implementation: A Dual Active Contour. IEEE Trans. Pattern Analysis and Machine Intelligence, 1997, 19(1): 63～68
[24] Williams D J, Shah M. A Fast algorithm for active contours and curvature estimation. CVGIP: Image Understanding, 1992, 55(1): 14～26
[25] 刘剑函，梁德群，王红光，等．一种具有形状约束的快速Snake类算法．中国图象图形学报，1999，4A(8)：644～648
[26] Mclnerney T, Terzopouios D. Topologically adaptable snakes. In Proceedings of 5th International Conference on Computer Vision, Cambridge, 1995:840～845
[27] Bhandarkar S M, Zhang Hui. Image Segmentation Using Evolutionary Computation. IEEE Trans. On Evolutionary Computation, 1999, 3(1): 1～21
[28] Hewer G A, Kenney C, Manjunath B S. Variational Image Segmentation Using Boundary Functions. IEEE Trans. On Image Processing, 1998, 7(9): 1269～1282
[29] Turk Mand, Pentland A. Eigenfaces for recognition. Journal Cognitive Neuroscience, 1991, 3:71～86
[30] Brunelli R, Poggio T. Face recognition: Feature versus Templates. IEEE Trans. On Pattern Analysis and Machine Intelligence, 1993, 15: 1042～1052
[31] 谌海新，沈振康，夏放怀．一种基于目标特征的多门限图像分割方法．电子学报，1999，27(3)：32～36
[32] Levitt T S, Hedgcock M W, Dye J W, et al. Bayesian inference for model-based segmentation of computed radiographs of the hand. Artificial Intelligence in Medicine, 1993,5(4): 365～387
[33] Nayar S K, Ikeuchi K, Kanade T. Suface Reflection: Physical and Geometrical Perspectives. IEEE Trans. On Pattern Analysis and Machine Intelligence, 1991, 13(7): 611～634
[34] Shafer S A. Using Color to Separate Reflection Components. Color Research and Application, 1985, 10: 210～218
[35] Klinker G J, Shafer S A, Kanade T. A Physical Approach to Color Image Understanding. International Joumal of Computer Vision, 1990, 4(1): 7～38
[36] Bajcsy R, Lee S W, Leonardis A. Color image segmentation with detection of highlights and local illumination induced by inter-reflection. In Proceedings of International Conference on Pattern Recognition, Atlantic City, N J, 1990:785～790
[37] Healey G. Using color for geometry-insensitive segmentation. Journal of the Optical Society of America, 1989, 6(6): 920～937
[38] Fabien Salzenstein, Wojciech Pieczynski. Paramater Ectimation in Hidden Fuzzy Markov Random Fields and Image Segmentation. Graphical Models and Image Processing, 1997, 59(4): 205～220
[39] Kapur T. Model Based Three Dimensional Medical Image Segmentation: [D], Artficial Intelligence Laboratory: Massachusetts Institute of Technology, 1999
[40] Dinstein I, Fong A C. Fast discrimination between homogeneous and textured region. In Proceedings of
7th Int. Conf. Pattern Recognition, 1984:361～362
[41] Uner M. Local Linear transforms for texture measurement. Signal Process, 1986,11:61～62
[42] Granlund G H. Description of texture using the general operator approach. In Proceedings of 5th Int. Conf. on Computer vision and Pattern Recognition, 1980:776～779
[43] Harwood D, Subbrao M, Davis L S. Texture classification by local rank correlation. CVGIP, 1985, 32: 404～405
[44] Pentland A P. Fractal-based description of natural scenes. IEEE PAMI, 1984, 6(6): 661～673
[45] Peleg J. Multiple resolution texture analysis and classification, IEEE PAMI, 1984, 6(4): 518～523
[46] 薛东辉．分形方法用于有噪声图像边缘检测得研究．通信学报，1996，7(1)：7～11
[47] Nguyen P T, Quinqueton J. Space filling curves and texture analysis. In Proceedings of 6th Int. Conf. on Pattern Recognition, 1982:282～283
[48] Derin H, Elliot H. Modeling and Segmentation of noisy and textured images using Gibbs random fields. IEEE PAMI, 1987, 9(1): 39～40
[49] 马小川．Markov模型及其在语音识别／图像分割中应用的研究：[博士学位论文]，西安：西北工业大学，1997
[50] Haralick R M. Statistical approaches to texture. In Proceedings of IEEE, 1979, 67(5): 786～804
[51] 刘宁宁，胡志刚，田捷，等．结合纹理信息的图像分割方法及其应用．软件学报，1999，10(4)：390～394
[2] Swain M.J. Ballard D.H. Indexing via color histograms. In proceedings of 3rd Int. Conf. Computer Vision, 1990:390～393
[3] 李向阳，鲁东明，潘云鹤．基于颜色的图像数据库检索方法的研究．计算机研究与发展，1999，3(6)：359～363
[4] Smith J.R. Integrated Spatial and Feature Image System: Retrieval, Analysis and Compression: [D]. New York: Columbia University, 1997
[5] Zhang Lei, Lin Fuzong, Zhang Bo. A CBIR Method Based on Color-Spatial Feature TENCON 99. In Proceedings of the 10th IEEE Region Conference, 1999, 1: 166～169
[6] Haralick R M, Shanmugam K D. Textural Features for Image Classification. IEEE Trans on SMC, 1973, 3: 610～621
[7] Carter P.H. Texture Discrimination Using Wavelets. In proceedings of the 1993 IEEE Computer Society Conference in Computer Vision and Pattern Recognition, 1993:640～641
[8] Ballard D.H. Generalizing the Hough Transform to Detect Arbitrary Shapes. Pattern Recognition, 1981, 13(2): 111～122
[9] Kass M, Witkin A, Terzopoulous D. Snakes: active contour models. In Proceedings of the 1 st International Conference on Computer Vision. London: IEEE Computer Society Press, 1987:259～268
[10] 柳伟，李国辉，曹莉华．一种基于内容的图象检索方法的实现．中国图象图形学报，1998，3(4)：304～308
[11] Amoid W M, Marce W, Simone S et al. Content-based image retrieval at the end of the early years. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2000, 22(12): 1349～1379.
[12] Hosoba M, et al. PACS Workstation for Computer Assisted Image Diagnosis. Processing SPIE Medical Imaging Conference Ⅳ, 1990
[13] 章毓晋．图像分割．北京：科学出版社，2001：2～4
[14] 章毓晋．图像分割．北京：科学出版社，2001：9～14
[15] Haralick R M. Digital step edges from zero crossing of second directional derivative. IEEE-PAMI, 1984, 6(1): 58～66
[16] Ostu N. A threshold selection method from gray-level histogram. IEEE Trans on SMC, 1979, 9:62～66
[17] O'gorman L, Sanderson A C. The converging square algorithm: an efficient method for locating peaks in multidimensions. IEEE-PAMI, 1984, 6:280～288
[18] O'gorman L, Sanderson A C. The wedge filter technique for convex boundary estimation. IEEE-PAMI, 1985, 7:326～332
[19] Hong T H. Rosenfeld A. Compact region extraction using weighted pixel linking in a pyramid. 1984, 6(2): 222～237
[20] Ron Schoenmakers. Integrated Methodology for Segmentation of Large Optical Satellite Images. In Land
Application of Remote Sensing. Luxembourg, Italy, 1995
[21] Kass M, Witkin A, Terzopoulos D. Snakes: Active Contour Models. Int. J. Computer Vision, 1988, 1(4): 321～331
[22] Chiou G I, Hwang J N. A Neural Network-based Stockhastic Active contour model (NNS-SNAKE) for contour finding of distinct features. IEEE Trans. On Medical Imaging, 1995, 4:1407～1416
[23] Gunn S R, Nixion M S. A Robust Snake Implementation: A Dual Active Contour. IEEE Trans. Pattern Analysis and Machine Intelligence, 1997, 19(1): 63～68
[24] Williams D J, Shah M. A Fast algorithm for active contours and curvature estimation. CVGIP: Image Understanding, 1992, 55(1): 14～26
[25] 刘剑函，梁德群，王红光，等．一种具有形状约束的快速Snake类算法．中国图象图形学报，1999，4A(8)：644～648
[26] Mclnerney T, Terzopouios D. Topologically adaptable snakes. In Proceedings of 5th International Conference on Computer Vision, Cambridge, 1995:840～845
[27] Bhandarkar S M, Zhang Hui. Image Segmentation Using Evolutionary Computation. IEEE Trans. On Evolutionary Computation, 1999, 3(1): 1～21
[28] Hewer G A, Kenney C, Manjunath B S. Variational Image Segmentation Using Boundary Functions. IEEE Trans. On Image Processing, 1998, 7(9): 1269～1282
[29] Turk Mand, Pentland A. Eigenfaces for recognition. Journal Cognitive Neuroscience, 1991, 3:71～86
[30] Brunelli R, Poggio T. Face recognition: Feature versus Templates. IEEE Trans. On Pattern Analysis and Machine Intelligence, 1993, 15: 1042～1052
[31] 谌海新，沈振康，夏放怀．一种基于目标特征的多门限图像分割方法．电子学报，1999，27(3)：32～36
[32] Levitt T S, Hedgcock M W, Dye J W, et al. Bayesian inference for model-based segmentation of computed radiographs of the hand. Artificial Intelligence in Medicine, 1993,5(4): 365～387
[33] Nayar S K, Ikeuchi K, Kanade T. Suface Reflection: Physical and Geometrical Perspectives. IEEE Trans. On Pattern Analysis and Machine Intelligence, 1991, 13(7): 611～634
[34] Shafer S A. Using Color to Separate Reflection Components. Color Research and Application, 1985, 10: 210～218
[35] Klinker G J, Shafer S A, Kanade T. A Physical Approach to Color Image Understanding. International Joumal of Computer Vision, 1990, 4(1): 7～38
[36] Bajcsy R, Lee S W, Leonardis A. Color image segmentation with detection of highlights and local illumination induced by inter-reflection. In Proceedings of International Conference on Pattern Recognition, Atlantic City, N J, 1990:785～790
[37] Healey G. Using color for geometry-insensitive segmentation. Journal of the Optical Society of America, 1989, 6(6): 920～937
[38] Fabien Salzenstein, Wojciech Pieczynski. Paramater Ectimation in Hidden Fuzzy Markov Random Fields and Image Segmentation. Graphical Models and Image Processing, 1997, 59(4): 205～220
[39] Kapur T. Model Based Three Dimensional Medical Image Segmentation: [D], Artficial Intelligence Laboratory: Massachusetts Institute of Technology, 1999
[40] Dinstein I, Fong A C. Fast discrimination between homogeneous and textured region. In Proceedings of
7th Int. Conf. Pattern Recognition, 1984:361～362
[41] Uner M. Local Linear transforms for texture measurement. Signal Process, 1986,11:61～62
[42] Granlund G H. Description of texture using the general operator approach. In Proceedings of 5th Int. Conf. on Computer vision and Pattern Recognition, 1980:776～779
[43] Harwood D, Subbrao M, Davis L S. Texture classification by local rank correlation. CVGIP, 1985, 32: 404～405
[44] Pentland A P. Fractal-based description of natural scenes. IEEE PAMI, 1984, 6(6): 661～673
[45] Peleg J. Multiple resolution texture analysis and classification, IEEE PAMI, 1984, 6(4): 518～523
[46] 薛东辉．分形方法用于有噪声图像边缘检测得研究．通信学报，1996，7(1)：7～11
[47] Nguyen P T, Quinqueton J. Space filling curves and texture analysis. In Proceedings of 6th Int. Conf. on Pattern Recognition, 1982:282～283
[48] Derin H, Elliot H. Modeling and Segmentation of noisy and textured images using Gibbs random fields. IEEE PAMI, 1987, 9(1): 39～40
[49] 马小川．Markov模型及其在语音识别／图像分割中应用的研究：[博士学位论文]，西安：西北工业大学，1997
[50] Haralick R M. Statistical approaches to texture. In Proceedings of IEEE, 1979, 67(5): 786～804
[51] 刘宁宁，胡志刚，田捷，等．结合纹理信息的图像分割方法及其应用．软件学报，1999，10(4)：390～394