| Инд. авторы: | Бериков В.Б., Пестунов И.А. |
| Заглавие: | Построение кластерного ансамбля для сегментации гиперспектральных изображений |
| Библ. ссылка: | Бериков В.Б., Пестунов И.А. Построение кластерного ансамбля для сегментации гиперспектральных изображений // Вычислительные технологии. - 2016. - Т.21. - № 1. - С.15-24. - ISSN 1560-7534. - EISSN 2313-691X. |
| Внешние системы: | РИНЦ: 25644522; |
| Реферат: | rus: Предложен алгоритм сегментации гиперспектральных изображений, основанный на коллективном подходе к кластерному анализу. Решение строится с помощью вычисления усредненной коассоциативной матрицы прототипов. Эффективность алгоритма исследуется на реальных гиперспектральных изображениях при наличии зашумленных каналов. eng: Ensemble approach has been actively developed in cluster analysis. This approach helps to reduce the dependence of the results on the choice of the algorithm parameters and to receive more stable solutions for noisy data. In this work we suggest an algorithm of hyperspectral images segmentation based on the ensemble clustering. For this purpose we consider a method of solution formation using co-association matrices that define how often pairs of objects appear in the same cluster in different variants of partitioning. One of the serious problems in constructing the ensemble solution is considerable running time of algorithms and necessity to store co-association matrices of large dimension in memory. Existing algorithms are not able to analyze large amounts of data, typical of hyperspectral images. In this paper we describe a computationally efficient algorithm for clustering ensemble. The main idea of the algorithm is based on the combination of data compression and ensemble clustering. While constructing ensemble solution one should examine not all pairs of observations, but rather only small number of pairs of “prototypes” that represent clusters. The effectiveness of the algorithm is illustrated on real hyperspectral images in the presence of noisy channels. It is shown that the proposed algorithm can improve the quality for the results of the noisy data analysis to handle a large images. |
| Ключевые слова: | гиперспектральное изображение; cluster analysis; ансамбль алгоритмов; коассоциативная матрица; кластерный анализ; Hyperspectral image; Co-association matrix; Ensemble algorithms; |
| Издано: | 2016 |
| Физ. характеристика: | с.15-24 |
| Цитирование: | 1. Bondur, V.G. Modern approaches for processing of big hyperspectral aerospace data. Issledovanie Zemli iz Kosmosa. 2014; 1:4-16. (In Russ.) 2. Schowengerdt, R.A. Remote sensing: models and methods for image processing. New York: Academic Press; 2006: 560. 3. Duda, R.O., Hart, P.E., Stork, D.G. Pattern classification. Second edition. New York: Wiley; 2000: 680. 4. Gonzales, R., Woods, R. Digital image processing. Third edition. New Jersey: Pearson Education Inc., Prentice Hall; 2008: 954. 5. Pestunov, I.A., Sinyavskiy, Yu.N. Сlustering algorithms in satellite images segmentation tasks. Bulletin of Kemerovo State University. 2012; 52(4/2):110-125. (In Russ.) 6. Pestunov, I.A., Rylov, S.A. Spectral-textural segmentation algorithms for satellite images with high spatial resolution. Bulletin of Kemerovo State University. 2012; 52(4/2): 104-110. (In Russ.) 7. Jain, A.K. Data clustering: 50 years beyond K-means. Pattern Recognition Letters. 2010; 31(8): 651-666. 8. Topchy, A.P., Law, M.H.C., Jain, A.K., Fred, A.L. Analysis of consensus partition in cluster Ensemble. Fourth IEEE International Conference on Data Mining (ICDM’04). 2004: 225-232. 9. Ghosh, J., Acharya, A. Cluster ensembles. Wiley Interdisciplinary Reviews: Data Mining and Knowledge Discovery. 2011; 1(5): 305-315. 10. Pestunov, I.A., Berikov, V.B., Sinyavskiy, Yu.N. Algorithm for multispectral image segmentation based оn ensemble of nonparametric clustering algorithms. Vestnik SibGAU. 2010; 5 (31): 56-64. (In Russ.) 11. Pestunov, I.A., Berikov, V.B., Kulikova, E.A., Rylov, S.A. Ensemble of clustering algorithm for large datasets. Optoelectronics, Instrumentation and Data Processing. 2011; 47(3): 245-252. 12. Pestunov, I.A., Rylov, S.A., Berikov, V.B. Hierarchical clustering algorithms for segmentation of multispectral images. Optoelectronics, Instrumentation and Data Processing. 2015; 51(4): 329-338. 13. Berikov, V. Weighted ensemble of algorithms for complex data clustering. Pattern Recognition Letters. 2014; (38): 99-106. 14. Arbelaitz, O., Gurrutxaga, I., Muguerza, J., Perez, J. , Perona, I. An extensive comparative study of cluster validity indices. Pattern Recognition Letters. 2013; 46(1): 243-256. 15. Hyperspectral Remote Sensing Scenes. Accessible at: http://www.ehu.eus/ccwintco/index.php? title=Hyperspectral_Remote_Sensing_Scenes (accessed at 20.11.2015). |
