分形方法在SAR图像区域分割中的应用
|
摘要
合成孔径雷达(Synthetic Aperture Radar，SAR)由于具有全天候、全天时、能穿透云雾、烟尘大面积地获取地表信息的特点，成为现代遥感技术研究的一个热点问题。随着国内外机载、星载SAR系统的发展，获取了大量的SAR图像数据，这些数据在当今信息时代发挥着日益重要的作用。由于SAR成像机理与传统的光学传感器有很大的差异，其数据处理的难度主要在于：SAR成像过程中电磁波与地表目标相互作用的机理还有待进一步揭示，人们难以从SAR图像直接反演地表物理现象；成像过程所必然带来的相干斑(Speckle)噪声使地物目标在SAR图像中具有独特的信息特点，常规的基于图像灰度的数字图像处理与分析技术难以取得令人满意的效果。与SAR数据源的快速增长趋势相比，对SAR图像处理与分析的理论和技术的研究相对滞后。作为获取地面信息的重要信息来源，SAR图像的解译技术是当前遥感以及计算机视觉领域的前沿课题。
SAR图像区域分割技术有助于对SAR图像中所含分布式目标区域的信息进行分析与解译。本文首先对SAR图像的统计特性和信息特点进行了深入的研究，得出结论：SAR图像以灰度变化的统计规律而不是以灰度值本身来反映地表目标的相关信息；SAR图像中分布式地物目标区域由于Speckle噪声的存在，更多地反映为一种具有一定粗糙度的纹理特性。借助纹理分类的概念，本论文主要研究了基于分形理论的SAR图像区域分割方法。分形方法是一种有效的纹理分析方法，利用图像的分形维数作为纹理特征进行区域分割。本文通过对分形维数特征空间采用结合图像边缘特征和迭代滤波的思想进行处理，提高了SAR图像区域分割的准确性。同时在文中提出了基于A／G系数的边缘检测算法，为区域分割提供了良好的区域边缘特征。
本论文最后对模拟图像和真实SAR图像的分割结果表明，所采用的分形方法对SAR图像的区域分割是行之有效的。
Synthetic Aperture Radar (SAR) plays an important role in modern remote sensing. It was broadly applied in almost every field of economy and society sustainable development because of its capability of providing all weather, full-time information about the earth. SAR images are very different from the visual or infrared images commonly used in remote sensing. The primary difficulties of SAR image processing and analysis lie both on that the interaction between electromagnetic wave and ground objects have not been revealed completely and that SAR image is subject to Speckle phenomenon which disturb the interpret of SAR images. The traditional digital image processing and analysis techniques based on grayscale that work successfully on nature images often do not perform as well on SAR images. With the rapid growth of available SAR data, the development of SAR image understanding techniques was desired. SAR image interpretation is now the urgent issue of remote sensing and computer vision.
Segmentation of SAR image based on texture is a critical preliminary operation in various SAR image processing application. This thesis focuses on the region segmentation of SAR images based on fractal method .The method is realized by exploiting the fractal dimension as the texture feature combined with edge character and iterative filter on the basis of the characteristics of the SAR images. Besides, an edge detection algorithm based on A/G coefficient is elaborated for improving efficiency of region edge detection, it affords a good edge character for region segmentation.
In the end of this paper we give some experiment results of synthetic images and real SAR images that show fractal method proposed is effective.
SAR图像区域分割技术有助于对SAR图像中所含分布式目标区域的信息进行分析与解译。本文首先对SAR图像的统计特性和信息特点进行了深入的研究，得出结论：SAR图像以灰度变化的统计规律而不是以灰度值本身来反映地表目标的相关信息；SAR图像中分布式地物目标区域由于Speckle噪声的存在，更多地反映为一种具有一定粗糙度的纹理特性。借助纹理分类的概念，本论文主要研究了基于分形理论的SAR图像区域分割方法。分形方法是一种有效的纹理分析方法，利用图像的分形维数作为纹理特征进行区域分割。本文通过对分形维数特征空间采用结合图像边缘特征和迭代滤波的思想进行处理，提高了SAR图像区域分割的准确性。同时在文中提出了基于A／G系数的边缘检测算法，为区域分割提供了良好的区域边缘特征。
本论文最后对模拟图像和真实SAR图像的分割结果表明，所采用的分形方法对SAR图像的区域分割是行之有效的。
Synthetic Aperture Radar (SAR) plays an important role in modern remote sensing. It was broadly applied in almost every field of economy and society sustainable development because of its capability of providing all weather, full-time information about the earth. SAR images are very different from the visual or infrared images commonly used in remote sensing. The primary difficulties of SAR image processing and analysis lie both on that the interaction between electromagnetic wave and ground objects have not been revealed completely and that SAR image is subject to Speckle phenomenon which disturb the interpret of SAR images. The traditional digital image processing and analysis techniques based on grayscale that work successfully on nature images often do not perform as well on SAR images. With the rapid growth of available SAR data, the development of SAR image understanding techniques was desired. SAR image interpretation is now the urgent issue of remote sensing and computer vision.
Segmentation of SAR image based on texture is a critical preliminary operation in various SAR image processing application. This thesis focuses on the region segmentation of SAR images based on fractal method .The method is realized by exploiting the fractal dimension as the texture feature combined with edge character and iterative filter on the basis of the characteristics of the SAR images. Besides, an edge detection algorithm based on A/G coefficient is elaborated for improving efficiency of region edge detection, it affords a good edge character for region segmentation.
In the end of this paper we give some experiment results of synthetic images and real SAR images that show fractal method proposed is effective.
引文
[1] Walessa M, Datcu M ,Model-Based Despeckling and Information Extraction from SAR Images, IEEE Trans. On Geoscience and Remote Sensing, Sept 2000, Vol. 38, No. 5, pp:2258-2269.
[2] 黄韦艮，周长宝，历冬玲，我国星载SAR海洋应用的现状与需求，中国航天，1997(12)，5-9．
[3] 章毓晋，图象分割，2001年，科学出版社．
[4] Simard M., Gianfranco DeGrandi, Keith P.B. Thomson, Goze B. Benie, Analysis of Speckle Noise Contribution on Wavelet Decomposition of SAR Images, IEEE Trans. On Geoscience and Remote Sensing, 1998, Vol. 36, No.6, pp:1953-1962.
[5] Zarrt A. E., Ziou D., Wang S., Jiang Q., Segmentation of SAR Images, The Journal of the Pattern Society, 2002, Vol. 35, pp:713-724.
[6] Fjortoft R., A. Lopes, P. Marthon, E. Cubero-Castan, An Optimal Multi-edge detector for SAR Image Segmentation, IEEE Trans. on Geoscience and Remote Sensing, 1998, Vol. 36, No. 3.
[7] Cook R., I. Mc. Connell, C. Oliver, MUM (Merge Using Moments) Segmentation for SAR Images, In Proc. SPIE ,1994, Vol. 2316, pp:92-103.
[8] Oliver C.J., S. Quegan. Understanding Synthetic Aperture Radar Images, 1998, Artech House Inc., 685 Canton Street, Norwood, MA 02062.
[9] Dong Y., Forester B. C., Segmentation of Radar Imagery Using the Gaussian Markov Random Field Model, International Journal of Remote Sensing, 1999, Vol. 20, No. 8, pp:1617-1639.
[10] Lemarechal C., R. Fjortoft, P. Marthon, E. Cubero-Castan, A. Lopes, SAR Image Segmentation by Morphological Methods, In Proc. SAR Image Analysis, Modelling and TechniQues Ⅲ, SPIE Vol. 3497, 1998.
[11] 赵峰，赵荣椿，纹理分割及特征提取方法综述，中国体视学与图象分析，1998年12月，第3卷，第4期，238-245．
[12] 杨绍国，尹忠科，罗炳伟，分形多尺度纹理图像分割，模式识别与人工智能，1998年，第11卷，第3期，347-351．
[13] 王润生，图像理解，1995年，国防科技大学出版社．
[14] J.W. Goodman, Statistical Properties of Laser Speckle Patterns, Laser Speckle and Related Phenomena, J.C. Dainty(ed.), New York: Springer-Verlag, 1984, pp:9-75.
[15] 刘永坦等著，雷达成像技术，1999年，哈尔滨工业大学出版社．
[16] 郭华东等著，雷达对地观测理论与应用，2000年，科学出版社．
[17] M. Walessa, Bayesian lnformation Extraction from SAR Image. Doctor Dissertation, 2001.
[18] R. Touzi, A Review of Speckle Filtering in the Context of Estimation Theory. IEEE Trans. Geoscience and Remote Sensing, Nov 2002, Vol. 40, NO. 11, pp:2392-2404.
[19] 管鲍，孙洪，李小玮，吕毅，SAR图像恢复的迭代方法，武汉大学学报(理学版)，2002年，第48卷第3期，355-359．
[20] F.N.S. Medeiros, N.D.A. Mascarenhas, da Costa, Adaptive Speckle MAP Filtering for SAR Images Using Statistical Clustering. Computer Graphics, Image Processing, and Vision, 1998. Proceedings. SIBGRAPI '98. International Symposium on, 1998.
[21] A. Lopes, E. Nezry, R. Touzi, H. Laur, Maximum A Posteriori Speckle Filtering And First Order Texture Models In Sar Images. Geoscience and Remote Sensing Symposium, 1990. IGARSS '90. 'Remote Sensing Science for the Nineties'., 10th Annual International, 1990.
[22] S. Foucher, G.B. Benie, J.M. Boucher, Multiscale MAP Filtering of SAR Images. IEEE Transactions on Image Processing, Jan 2001, Vol. 10, pp: 49-60.
[23] 徐新，星载SAR图像处理与分析中若干关键技术的研究，武汉大学博士学位论文，编号：10486，2003．
[24] 张济忠，分形，1995，清华大学出版社．
[25] 高安秀树著，沈步明，常子文译，分数维，1994，地震出版社．
[26] Keller R,R. Crownover, R. chen, Characteristics of Neural Scenes Related to the Fractal Dimension, IEEE Trans. PAMI-9, (5) 1987, pp:621-627.
[27] Nirupam Sarkar ,B.B. Chaudhuri, An Efficient Approach to Estimate Fractal Dimension of Textural Images, Pattern Recognition, 1992, Vol. 25, No. 9, pp:661-674.
[28] N. Yokoya, K. Yamamoto, Fractal-Based Analysis and Interpolation of 3-D Natural Surface Shapes and their Application to Terrain Modeling, Computer Vision, Graphics and Images Processing ,1989, Vol. 46, pp:284-302.
[29] Therrien C W, An Estimation-Theoretic Approach to Terrain Image Segmentation, Computer Vision, Graphics and Images Processing, 1983, Vol. 22, No. 3, pp:313-326.
[30] N. Sarkar , B.B. Chaudhuri, An Efficient Differential Box-Counting Approach to Compute the Fractal Dimension of Image. IEEE Trans. Systems Man Cybern, 1994, Vol. 24, pp:115-120.
[31] Ulaby, F.T.,R.K. Moore, A.K. Fung, Microware Remote Sensing, Active and Passive, Vo. Ⅱ: Radar Remote Sensing and Surface Scattering and Emission
Theory, Addision-Wesley, Reading, MA, MA, 457-1064.
[32] 舒士畏，洪峻，林兵，目标的雷达图像特征及其提取方法研究，雷达图像分析及地质应用，1991，科学出版社．
[33] 刘浩，郭华东，雷达图像纹理信息的提取及在地学分析中的应用，环境遥感，1995．
[34] 王东生，曹磊，混沌、分形及其应用，1995，中国科学技术大学出版社．
[35] B.B. Chaudhuri, Nirupam Sakar, Texture Segmentation Using Fractal Dimension, IEEE Transaction on Pattern Analysis and Machine Intelligence, Jan 1995, Vol. 17, No. 1, pp:72-77.
[36] 吴更石，梁德群，田园，基于分形维数的的纹理图像分割，计算机学报，1999年，第22卷，第10期，1109-1113．
[37] Touzi R, Lopes A, Bousquet P, A Statistical and Geometrical Edge Detector for SAR Images , IEEE Transactions on Geosicense and Remote Sensing, 1988, Vol. 26, No. 6, pp:764-773.
[38] Iain H. Woodhouse, The Ratio of the Arith- metic to the Geometric Mean: A Cross-Entropy Interpretation , IEEE Transaction on Geosicence and Remote Sensing, 2001, Vol. 39, No. 1, pp:188-189.
[39] 李小玮，基于统计处理的合成孔径雷达图像滤波方法研究．，武汉大学硕士毕业论文，2002．
[40] 杨枝灵，王开等著，VisualC++数字图像获取、处理及实践应用，2003年，人民邮电出版社．
[41] A.P. Pentland, Fractal-Based Description of Natural Scenses, IEEE Transaction on Pattern Analysis and Machine Intelligence ,1984, Vol. PAMI-6, No. 6.
[42] J.M. Keller, S. Chen, Texture Description and Segmentation through Fractal Geometry ,Computer Vision, Graphics and Images Processing ,1989, Vol. 45, pp: 150-166.
[43] P.P. Ohanlan ,R.C. Dubes, Performance Evaluation for four Classes of Textural Feature ,Pattern Recognition, 1992, Vol. 25, No. 8, pp:819-833.
[44] C.V. Stewart, Fractional Brownian Motion Models for SAR Imagery Scene Segmentation, Proceedings of IEEE, 1993, Vol. 81, No. 10, pp:1511-1521.
[45] G.M.Henebry, H.Kux, Lacunarityas a TextureMeasurefor SARImagery,International Journal on Remote Sensing,1995, Vol.16,No.3,pp:565-571.
[2] 黄韦艮，周长宝，历冬玲，我国星载SAR海洋应用的现状与需求，中国航天，1997(12)，5-9．
[3] 章毓晋，图象分割，2001年，科学出版社．
[4] Simard M., Gianfranco DeGrandi, Keith P.B. Thomson, Goze B. Benie, Analysis of Speckle Noise Contribution on Wavelet Decomposition of SAR Images, IEEE Trans. On Geoscience and Remote Sensing, 1998, Vol. 36, No.6, pp:1953-1962.
[5] Zarrt A. E., Ziou D., Wang S., Jiang Q., Segmentation of SAR Images, The Journal of the Pattern Society, 2002, Vol. 35, pp:713-724.
[6] Fjortoft R., A. Lopes, P. Marthon, E. Cubero-Castan, An Optimal Multi-edge detector for SAR Image Segmentation, IEEE Trans. on Geoscience and Remote Sensing, 1998, Vol. 36, No. 3.
[7] Cook R., I. Mc. Connell, C. Oliver, MUM (Merge Using Moments) Segmentation for SAR Images, In Proc. SPIE ,1994, Vol. 2316, pp:92-103.
[8] Oliver C.J., S. Quegan. Understanding Synthetic Aperture Radar Images, 1998, Artech House Inc., 685 Canton Street, Norwood, MA 02062.
[9] Dong Y., Forester B. C., Segmentation of Radar Imagery Using the Gaussian Markov Random Field Model, International Journal of Remote Sensing, 1999, Vol. 20, No. 8, pp:1617-1639.
[10] Lemarechal C., R. Fjortoft, P. Marthon, E. Cubero-Castan, A. Lopes, SAR Image Segmentation by Morphological Methods, In Proc. SAR Image Analysis, Modelling and TechniQues Ⅲ, SPIE Vol. 3497, 1998.
[11] 赵峰，赵荣椿，纹理分割及特征提取方法综述，中国体视学与图象分析，1998年12月，第3卷，第4期，238-245．
[12] 杨绍国，尹忠科，罗炳伟，分形多尺度纹理图像分割，模式识别与人工智能，1998年，第11卷，第3期，347-351．
[13] 王润生，图像理解，1995年，国防科技大学出版社．
[14] J.W. Goodman, Statistical Properties of Laser Speckle Patterns, Laser Speckle and Related Phenomena, J.C. Dainty(ed.), New York: Springer-Verlag, 1984, pp:9-75.
[15] 刘永坦等著，雷达成像技术，1999年，哈尔滨工业大学出版社．
[16] 郭华东等著，雷达对地观测理论与应用，2000年，科学出版社．
[17] M. Walessa, Bayesian lnformation Extraction from SAR Image. Doctor Dissertation, 2001.
[18] R. Touzi, A Review of Speckle Filtering in the Context of Estimation Theory. IEEE Trans. Geoscience and Remote Sensing, Nov 2002, Vol. 40, NO. 11, pp:2392-2404.
[19] 管鲍，孙洪，李小玮，吕毅，SAR图像恢复的迭代方法，武汉大学学报(理学版)，2002年，第48卷第3期，355-359．
[20] F.N.S. Medeiros, N.D.A. Mascarenhas, da Costa, Adaptive Speckle MAP Filtering for SAR Images Using Statistical Clustering. Computer Graphics, Image Processing, and Vision, 1998. Proceedings. SIBGRAPI '98. International Symposium on, 1998.
[21] A. Lopes, E. Nezry, R. Touzi, H. Laur, Maximum A Posteriori Speckle Filtering And First Order Texture Models In Sar Images. Geoscience and Remote Sensing Symposium, 1990. IGARSS '90. 'Remote Sensing Science for the Nineties'., 10th Annual International, 1990.
[22] S. Foucher, G.B. Benie, J.M. Boucher, Multiscale MAP Filtering of SAR Images. IEEE Transactions on Image Processing, Jan 2001, Vol. 10, pp: 49-60.
[23] 徐新，星载SAR图像处理与分析中若干关键技术的研究，武汉大学博士学位论文，编号：10486，2003．
[24] 张济忠，分形，1995，清华大学出版社．
[25] 高安秀树著，沈步明，常子文译，分数维，1994，地震出版社．
[26] Keller R,R. Crownover, R. chen, Characteristics of Neural Scenes Related to the Fractal Dimension, IEEE Trans. PAMI-9, (5) 1987, pp:621-627.
[27] Nirupam Sarkar ,B.B. Chaudhuri, An Efficient Approach to Estimate Fractal Dimension of Textural Images, Pattern Recognition, 1992, Vol. 25, No. 9, pp:661-674.
[28] N. Yokoya, K. Yamamoto, Fractal-Based Analysis and Interpolation of 3-D Natural Surface Shapes and their Application to Terrain Modeling, Computer Vision, Graphics and Images Processing ,1989, Vol. 46, pp:284-302.
[29] Therrien C W, An Estimation-Theoretic Approach to Terrain Image Segmentation, Computer Vision, Graphics and Images Processing, 1983, Vol. 22, No. 3, pp:313-326.
[30] N. Sarkar , B.B. Chaudhuri, An Efficient Differential Box-Counting Approach to Compute the Fractal Dimension of Image. IEEE Trans. Systems Man Cybern, 1994, Vol. 24, pp:115-120.
[31] Ulaby, F.T.,R.K. Moore, A.K. Fung, Microware Remote Sensing, Active and Passive, Vo. Ⅱ: Radar Remote Sensing and Surface Scattering and Emission
Theory, Addision-Wesley, Reading, MA, MA, 457-1064.
[32] 舒士畏，洪峻，林兵，目标的雷达图像特征及其提取方法研究，雷达图像分析及地质应用，1991，科学出版社．
[33] 刘浩，郭华东，雷达图像纹理信息的提取及在地学分析中的应用，环境遥感，1995．
[34] 王东生，曹磊，混沌、分形及其应用，1995，中国科学技术大学出版社．
[35] B.B. Chaudhuri, Nirupam Sakar, Texture Segmentation Using Fractal Dimension, IEEE Transaction on Pattern Analysis and Machine Intelligence, Jan 1995, Vol. 17, No. 1, pp:72-77.
[36] 吴更石，梁德群，田园，基于分形维数的的纹理图像分割，计算机学报，1999年，第22卷，第10期，1109-1113．
[37] Touzi R, Lopes A, Bousquet P, A Statistical and Geometrical Edge Detector for SAR Images , IEEE Transactions on Geosicense and Remote Sensing, 1988, Vol. 26, No. 6, pp:764-773.
[38] Iain H. Woodhouse, The Ratio of the Arith- metic to the Geometric Mean: A Cross-Entropy Interpretation , IEEE Transaction on Geosicence and Remote Sensing, 2001, Vol. 39, No. 1, pp:188-189.
[39] 李小玮，基于统计处理的合成孔径雷达图像滤波方法研究．，武汉大学硕士毕业论文，2002．
[40] 杨枝灵，王开等著，VisualC++数字图像获取、处理及实践应用，2003年，人民邮电出版社．
[41] A.P. Pentland, Fractal-Based Description of Natural Scenses, IEEE Transaction on Pattern Analysis and Machine Intelligence ,1984, Vol. PAMI-6, No. 6.
[42] J.M. Keller, S. Chen, Texture Description and Segmentation through Fractal Geometry ,Computer Vision, Graphics and Images Processing ,1989, Vol. 45, pp: 150-166.
[43] P.P. Ohanlan ,R.C. Dubes, Performance Evaluation for four Classes of Textural Feature ,Pattern Recognition, 1992, Vol. 25, No. 8, pp:819-833.
[44] C.V. Stewart, Fractional Brownian Motion Models for SAR Imagery Scene Segmentation, Proceedings of IEEE, 1993, Vol. 81, No. 10, pp:1511-1521.
[45] G.M.Henebry, H.Kux, Lacunarityas a TextureMeasurefor SARImagery,International Journal on Remote Sensing,1995, Vol.16,No.3,pp:565-571.