روشی برای بخش بندی تصاویر سنجش از دور با استفاده از الگوریتم Watershed و خوشه بندی Fuzzy C-Means
محورهای موضوعی : مهندسی الکترونیک
محسن حامد
1
(آموزشکده فنی و حرفه ای سما، دانشگاه آزاد اسلامی، واحد بوشهر، بوشهر، ایران)
فاطمه حاجیانی
2
(دانشگاه آزاد اسلامی واحد خورموج، ایران، عضو هیات علمی)
کلید واژه: Segmentation, سنجش از دور, remote sensing, Watershed Algorithm, بخش بندی, الگوریتمWatershed, خوشه بندی Fuzzy C-Means, Fuzzy C-Means clustering,
چکیده مقاله :
در تقسیم بندی پیکسل های تصویر سنجش از دور با استفاده از بخش بندی Watershed، مرزهای تصویر به خوبی آشکار نمی شوند. در این مقاله، یک الگوریتم خوشه بندی تصویر بر اساس بخش بندی Watershed و خوشه بندی Fuzzy C-Means ارائه شده است. روش کار به این صورت است که در ابتدا از الگوریتم Watershed برای بخش بندی تصویر حاصل از مجموع مشتق تصویر با تصویر اصلی استفاده می شود. مشتق گرفتن از تصویر موجب می شود مرزهای تصویر به خوبی آشکار شده و رویهم افتادگی بین مرزها رخ ندهد. پس از بخش بندی، برای ترکیب نواحی مشابه حاصل، از خوشه بندی Fuzzy C-Means استفاده می شود. در نهایت، به منظور بهبود نتایج خوشه بندی، یک ماتریس تقسیم بندی جدید نیز برای هر ناحیه از تصویر، با توجه به ویژگی های نواحی همسایه ی آن محاسبه می شود. با توجه به اینکه تصاویر سنجش از دور شامل یک سطح نویز بالا هستند، الگوریتم پیشنهادی در مقایسه با الگوریتم Watershed رایج، توانایی بیشتری در مقابله با نویز دارد و لبه های تصویر بهتر نمایان می شوند. نتایج آزمایش روش پیشنهادی بر روی یک نمونه تصویر سنجش از دور، عملی بودن و کارایی الگوریتم پیشنهادی را نشان می دهد.
In the division of remote sensing image pixels using Watershed segmentation, the boundaries of the image are not well defined. In this paper, an image clustering algorithm based on Watershed segmentation and Fuzzy C-Means clustering is presented. The method is that first the Watershed algorithm is used to segment the image obtained from the sum of the image derivative with the original image. Image derivation makes the borders of the image well defined and does not overlap between the borders. After segmentation, Fuzzy C-Means clustering is used to combine similar regions. Finally, in order to improve the clustering results, a new segmentation matrix is calculated for each area of the image, according to the characteristics of its neighboring areas. Due to the fact that remote sensing images contain a high level of noise, the proposed algorithm is more capable of dealing with noise compared to the conventional Watershed algorithm, and the edges of the image appear better. The test results of the proposed method on a sample of remote sensing image show the practicality and efficiency of the proposed algorithm.
[1] M. Williamson, “Cambridge Dictionary of Space Technology,” Cambridge University Press, First Edition, 2001.
[2] J. C. Bezdek, “Pattern recognition with fuzzy objective function algorithms,” Plenum, New York, 1981.
[3] Zeng. Xuexing, Ren. Jinchang, McKee. David, Lavender. Samantha, Marshall. Stephen, "Pixel clustering and hyperspectral image segmentation for ocean colour remote sensing", In Hyperspectral Imaging Conference, 2012.
[4] Du Gen-yuan, Miao Fang; Sheng-Li, Tian; Liu Ye, "A Modified Fuzzy C-means Algorithm in Remote Sensing Image Segmentation", International Conference on Environmental Science and Information Application Technology, 4-5 July 2009, Vol.3, Pages 447–450.
[5] Li Ma, R.C. Staunton, “A modified fuzzy c-means image segmentation algorithm for use with uneven illumination patterns,” In Pattern Recognition Conference, 2007, Vol.. 40, No.11, pp.3005-3011.
[6] Leonid P. Podenok, RaufKh. Sadykhov, “Multispectral Satellite Image Segmentation Using Fuzzy Clustering and Nonlinear Filtering Methods,” International Machine Vision and Image Processing Conference, 2008, pp. 43-48.
[7] J. Fan, M. Han, and J. Wang, “Single point iterative weighted fuzzy c-means clustering algorithm for remote sensing image segmentation,” Pattern Recognition, 42(11), 2009, pp. 2527-2540.
[8] Leonid P. Podenok, RaufKh. Sadykhov, “Multispectral Satellite Image Segmentation Using Fuzzy Clustering and Nonlinear Filtering Methods,” International Machine Vision and Image Processing Conference, 2008, pp. 43-48.
[9] S. Saha and S. Bandyopadhyay, “A Symmetry Based Multiobjective Clustering Technique for Automatic Evolution of Clusters,” Pattern Recognition, vol. 43, No. 3, March 2010, pp. 738-751.
[10] M. Hasanzadeh, and s. kasaei, “A Multispectral Image Segmentation Method using Size-Weighted Fuzzy Clustering and Membership Connectedness, ” IEEE Geosc. and Remote Sen. Letters, Vol. 7, Issue 3, pp. 520-524, 2010.
[11] W. S. Aldrich, M. E. Kappus, and R. G. Resmini, “HYDICE post-flight data processing,” Proc. SPIE, vol. 2758, pp. 354–363, 1996.
[12] Pal. N. R, and Pal, S. K, “A Review on Image Segmentation Techniques,”Pattern Recognition, 26(9), p.p. 1274-1294, 1993.
[13] L. Vincent and P. Soille, “Watersheds in digital spaces: An efficient algorithm based on immersion simulations,” IEEE Trans. Pattern Anal. Machine Intel.vol. 13, pp. 583–598, 1991.
[14] Tarabalka. Y, Chanussot. J, and Benediktsson. J. A, “Segmentation and classification of hyperspectral images using watershed transformation”, Pattern Recognition, vol. 43, no. 7, pp. 2367-2379, July 2010.
[15] Beucher. S, “The watershed transformation applied to image Segmentation,” Scanning Microsc, Vol 6, p.p. 299-314, 2005.
[16] J. C. Bezdek, “Pattern recognition with fuzzy objective function algorithms,” Plenum, New York, 1981.
[17] J. C. Bezdek. “Cluster validity with fuzzy sets” Cybernetics and Systems, 1973, 3(3): pp. 58–73.
[18] J. C. Bezdek. “Mathematical models for systematic and taxonomy”. Proc. of Eight International Conference on NumericalTaxonomy, San Francisco, 1975: 143–166.
[19] R. Babuska, P. J. v. d. V. and Kaymak, U. (2002). “Improved covariance estimation for gustafson-kessel clustering,” In IEEE International Conference on Fuzzy Systems, pages 1081–1085.
[20] Gath, I. and Geva, A. B. (1989). “Unsupervised optimal fuzzy clustering.” In IEEE Transactions on Pattern Analysis and Machine Intelligence, pages 7:773–781.
_||_[1] M. Williamson, “Cambridge Dictionary of Space Technology,” Cambridge University Press, First Edition, 2001.
[2] J. C. Bezdek, “Pattern recognition with fuzzy objective function algorithms,” Plenum, New York, 1981.
[3] Zeng. Xuexing, Ren. Jinchang, McKee. David, Lavender. Samantha, Marshall. Stephen, "Pixel clustering and hyperspectral image segmentation for ocean colour remote sensing", In Hyperspectral Imaging Conference, 2012.
[4] Du Gen-yuan, Miao Fang; Sheng-Li, Tian; Liu Ye, "A Modified Fuzzy C-means Algorithm in Remote Sensing Image Segmentation", International Conference on Environmental Science and Information Application Technology, 4-5 July 2009, Vol.3, Pages 447–450.
[5] Li Ma, R.C. Staunton, “A modified fuzzy c-means image segmentation algorithm for use with uneven illumination patterns,” In Pattern Recognition Conference, 2007, Vol.. 40, No.11, pp.3005-3011.
[6] Leonid P. Podenok, RaufKh. Sadykhov, “Multispectral Satellite Image Segmentation Using Fuzzy Clustering and Nonlinear Filtering Methods,” International Machine Vision and Image Processing Conference, 2008, pp. 43-48.
[7] J. Fan, M. Han, and J. Wang, “Single point iterative weighted fuzzy c-means clustering algorithm for remote sensing image segmentation,” Pattern Recognition, 42(11), 2009, pp. 2527-2540.
[8] Leonid P. Podenok, RaufKh. Sadykhov, “Multispectral Satellite Image Segmentation Using Fuzzy Clustering and Nonlinear Filtering Methods,” International Machine Vision and Image Processing Conference, 2008, pp. 43-48.
[9] S. Saha and S. Bandyopadhyay, “A Symmetry Based Multiobjective Clustering Technique for Automatic Evolution of Clusters,” Pattern Recognition, vol. 43, No. 3, March 2010, pp. 738-751.
[10] M. Hasanzadeh, and s. kasaei, “A Multispectral Image Segmentation Method using Size-Weighted Fuzzy Clustering and Membership Connectedness, ” IEEE Geosc. and Remote Sen. Letters, Vol. 7, Issue 3, pp. 520-524, 2010.
[11] W. S. Aldrich, M. E. Kappus, and R. G. Resmini, “HYDICE post-flight data processing,” Proc. SPIE, vol. 2758, pp. 354–363, 1996.
[12] Pal. N. R, and Pal, S. K, “A Review on Image Segmentation Techniques,”Pattern Recognition, 26(9), p.p. 1274-1294, 1993.
[13] L. Vincent and P. Soille, “Watersheds in digital spaces: An efficient algorithm based on immersion simulations,” IEEE Trans. Pattern Anal. Machine Intel.vol. 13, pp. 583–598, 1991.
[14] Tarabalka. Y, Chanussot. J, and Benediktsson. J. A, “Segmentation and classification of hyperspectral images using watershed transformation”, Pattern Recognition, vol. 43, no. 7, pp. 2367-2379, July 2010.
[15] Beucher. S, “The watershed transformation applied to image Segmentation,” Scanning Microsc, Vol 6, p.p. 299-314, 2005.
[16] J. C. Bezdek, “Pattern recognition with fuzzy objective function algorithms,” Plenum, New York, 1981.
[17] J. C. Bezdek. “Cluster validity with fuzzy sets” Cybernetics and Systems, 1973, 3(3): pp. 58–73.
[18] J. C. Bezdek. “Mathematical models for systematic and taxonomy”. Proc. of Eight International Conference on NumericalTaxonomy, San Francisco, 1975: 143–166.
[19] R. Babuska, P. J. v. d. V. and Kaymak, U. (2002). “Improved covariance estimation for gustafson-kessel clustering,” In IEEE International Conference on Fuzzy Systems, pages 1081–1085.
[20] Gath, I. and Geva, A. B. (1989). “Unsupervised optimal fuzzy clustering.” In IEEE Transactions on Pattern Analysis and Machine Intelligence, pages 7:773–781.