基于模糊c均值算法和人工蜂群算法的无监督波段选择
|
摘要
波段选择是高光谱影像处理中一种重要的降维方法.在类标签不可获得的情况下,如何选择出一个具有代表性的波段子集是一个挑战性的问题.为了解决高光谱数据维数灾难以及光谱空间冗余的问题,基于模糊C均值算法(Fuzzy c-means,FCM),人工蜂群算法(Artificial Bee Colony, ABC)与极大熵准则(Maximum Entropy, ME),文章提出了一种新的无监督波段选择方法.该方法首先通过FCM算法将相似的波段划分到一个波段子集中,然后以ME为ABC算法中的适应度函数,寻找优化的波段子集.为验证该算法的有效性,在三个典型的高光谱数据集上,将所提出的方法和其它一些有效的波段选择算法进行了分类精度和计算时间对比.实验结果表明,所提出的算法不但可以得到高的分类精度,同时在计算时间上也具有明显的优势.
Band selection is an important dimension reduction technique in hyperspectral image processing. Under the condition that class labels are unavailable, how to select a representative band subset becomes a severe challenge. To address the problem of dimension disaster and spectral space redundancy, a novel unsupervised band selection method is introduced based on fuzzy c-means algorithm(FCM), artificial bee colony algorithm(ABC) and maximum Entropy(ME). In this method, the similar bands are firstly divided into a band subset by FCM algorithm, and then ME is adopted as the fitness function in the ABC algorithm to find the optimized band subset. To verify the effectiveness of the proposal, the proposed method is compared with other efficient band selection algorithms on three typical hyperspectral datasets.Experimental results show that the proposed algorithm not only can achieve high classification accuracy, but also has obvious advantages in computing time.
Band selection is an important dimension reduction technique in hyperspectral image processing. Under the condition that class labels are unavailable, how to select a representative band subset becomes a severe challenge. To address the problem of dimension disaster and spectral space redundancy, a novel unsupervised band selection method is introduced based on fuzzy c-means algorithm(FCM), artificial bee colony algorithm(ABC) and maximum Entropy(ME). In this method, the similar bands are firstly divided into a band subset by FCM algorithm, and then ME is adopted as the fitness function in the ABC algorithm to find the optimized band subset. To verify the effectiveness of the proposal, the proposed method is compared with other efficient band selection algorithms on three typical hyperspectral datasets.Experimental results show that the proposed algorithm not only can achieve high classification accuracy, but also has obvious advantages in computing time.
引文
[1]樊利恒,吕俊伟,邓江生,等.基于分类器集成的高光谱遥感图像分类方法.光学学报,2014, 34(9):91-101.(Fan L H, Lii J W, Deng J S, et al. Classification of hyperspectral remote sensing images based on bands grouping and classification ensembles. Acta Optica Sinica, 2014, 34(9):91-101.)
[2]谢娟英,谢维信.基于特征子集区分度与支持向量机的特征选择算法.计算机学报,2014, 37(8):1704-1718.(Xie J Y, Xie W X, et al. Several feature selection algorithms based on the discernibility of a feature subset and support vector machines. Chinese Journal of Computers, 2014, 37(8):1704-1718.)
[3]杜培军,王小美,谭琨,等.利用流形学习进行高光谱遥感影像的降维与特征提取.武汉大学学报(信息科学版), 2011, 36(2):148-152.(Du P J, Wang X M, Tan K, et al. Dimensionality reduction and feature extraction from hyperspectral remote sensing imagery based on manifold learning. Geomatics and Information Science of Wuhan University, 2011, 36(2):148-152.)
[4] Feng J, Jiao L, Liu F, et al. Unsupervised feature selection based on maximum information and minimum redundancy for hyperspectral images. Pattern Recognition, 2016, 51:295-309.
[5] Solorio-Fernandez S, Carrasco-Ochoa J A, Martinez-Trinidad J F. Hybrid feature selection method for supervised classification based on Laplacian score ranking, advances in pattern recognition—Second mexican conference on pattern, Puebla, Mexico, September, 2010.
[6]唐贵华.基于密度排序聚类和超像素分割的高光谱遥感影像降维方法研究.硕士论文,深圳大学,2016.(Tang G H. Ranking-based-clustering and superpixel segmentation for hyperspectral remote imagery dimensionality reduction. Master's thesis, Shenzhen University, 2016.)
[7] Zhang Y, Desai M D, Zhang J, et al. Adaptive subspace decomposition for hyperspectral data dimensionality reduction. International conference on Image Processing, IEEE, 1999, 2:326-329.
[8]王立国,赵亮,刘丹凤.基于人工蜂群算法高光谱图像波段选择.哈尔滨工业大学学报,2015,47(11):82-88.(Wang L G, Zhao L, Liu D F, et al. Artificial bee colony algorithm-based band selection for hyperspectral imagery. Journal of Harbin Institute of Technology, 2015, 47(11):82-88.)
[9]赵冬,赵光恒.基于改进遗传算法的高光谱图像波段选择.中国科学院大学学报,2009, 26(6):795-802.(Zhao D, Zhao G H. Band selection of hyperspectral image based on improved genetic algorithm.Journal of University of Chinese Academy of Sciences, 2009, 26(6):795-802.)
[10]王立国,魏芳洁.结合遗传算法和蚁群算法的高光谱图像波段选择.中国图象图形学报,2013, 18(2):235-242.(Wang L G, Wei F J. Band selection for hyperspectral imagery based on combination of genetic algorithm and ant colony algorithm. Journal of Image and Graphics, 2013, 18(2):235-242.)
[11] Ghamisi P, Benediktsson J. A. Feature selection based on hybridization of genetic algorithm and particle swarm optimization. IEEE Transactions on Geoscience Remote Sensing Letter, 2015, 12:309-313.
[12] Karaboga D, Basturk B A. powerful and efficient algorithm for numerical function optimization:Artificial bee colony(ABC)algorithm. Journal of Global Optimization, 2007, 39:459-471.
[13] Karaboga D, Basturk B. On the performance of artificial bee colony(ABC)algorithm. Applied Soft Computation, 2008, 8:687-697.
[14] Karaboga D, Ozturk C. A novel clustering approach:Artificial Bee Colony(ABC)algorithm.Applied Soft Computation, 2011, 11:652-657.
[15] Feng J, Jiao L. Zhang X, et al. Hyperspectral band selection based on trivariate mutual information and clonal selection. IEEE Transactions on Geoscience Remote Sensing, 2014, 52(7):4092-4105.
[16] Feng J, Jiao L, Liu F, et al. Mutual-information-based semi-supervised hyperspectral band selection with high discrimination, high information, and low redundancy. IEEE Transactions on Geoscience Remote Sensing, 2015, 53(5):2956-2969.
[17] Mitra P, Murthy C A, Pal S K. Unsupervised feature selection using feature similarity. IEEE Transactions on Pattern Analysis Machine Intelligence, 2002, 24(3):301-312.
[18] Wu J. Unsupervised intrusion feature selection based on genetic algorithm and FCM. Lecture Notes in Electrical Engineering, 2012, 154:1005-1012.
[19] Martinez-Usómartinez-Uso A, Pla F, Sotoca J M, et al. Clustering-based hyperspectral band selection using information measures. IEEE Transactions on Geoscience Remote Sensing, 2007,45(12):4158-4171.
[20] Bazi Y, Melgani F. Toward an optimal SVM classification system for hyperspectral remote sensing images. IEEE Transactions on Geoscience Remote Sensing, 2006, 44(11):3374-3385.
[2]谢娟英,谢维信.基于特征子集区分度与支持向量机的特征选择算法.计算机学报,2014, 37(8):1704-1718.(Xie J Y, Xie W X, et al. Several feature selection algorithms based on the discernibility of a feature subset and support vector machines. Chinese Journal of Computers, 2014, 37(8):1704-1718.)
[3]杜培军,王小美,谭琨,等.利用流形学习进行高光谱遥感影像的降维与特征提取.武汉大学学报(信息科学版), 2011, 36(2):148-152.(Du P J, Wang X M, Tan K, et al. Dimensionality reduction and feature extraction from hyperspectral remote sensing imagery based on manifold learning. Geomatics and Information Science of Wuhan University, 2011, 36(2):148-152.)
[4] Feng J, Jiao L, Liu F, et al. Unsupervised feature selection based on maximum information and minimum redundancy for hyperspectral images. Pattern Recognition, 2016, 51:295-309.
[5] Solorio-Fernandez S, Carrasco-Ochoa J A, Martinez-Trinidad J F. Hybrid feature selection method for supervised classification based on Laplacian score ranking, advances in pattern recognition—Second mexican conference on pattern, Puebla, Mexico, September, 2010.
[6]唐贵华.基于密度排序聚类和超像素分割的高光谱遥感影像降维方法研究.硕士论文,深圳大学,2016.(Tang G H. Ranking-based-clustering and superpixel segmentation for hyperspectral remote imagery dimensionality reduction. Master's thesis, Shenzhen University, 2016.)
[7] Zhang Y, Desai M D, Zhang J, et al. Adaptive subspace decomposition for hyperspectral data dimensionality reduction. International conference on Image Processing, IEEE, 1999, 2:326-329.
[8]王立国,赵亮,刘丹凤.基于人工蜂群算法高光谱图像波段选择.哈尔滨工业大学学报,2015,47(11):82-88.(Wang L G, Zhao L, Liu D F, et al. Artificial bee colony algorithm-based band selection for hyperspectral imagery. Journal of Harbin Institute of Technology, 2015, 47(11):82-88.)
[9]赵冬,赵光恒.基于改进遗传算法的高光谱图像波段选择.中国科学院大学学报,2009, 26(6):795-802.(Zhao D, Zhao G H. Band selection of hyperspectral image based on improved genetic algorithm.Journal of University of Chinese Academy of Sciences, 2009, 26(6):795-802.)
[10]王立国,魏芳洁.结合遗传算法和蚁群算法的高光谱图像波段选择.中国图象图形学报,2013, 18(2):235-242.(Wang L G, Wei F J. Band selection for hyperspectral imagery based on combination of genetic algorithm and ant colony algorithm. Journal of Image and Graphics, 2013, 18(2):235-242.)
[11] Ghamisi P, Benediktsson J. A. Feature selection based on hybridization of genetic algorithm and particle swarm optimization. IEEE Transactions on Geoscience Remote Sensing Letter, 2015, 12:309-313.
[12] Karaboga D, Basturk B A. powerful and efficient algorithm for numerical function optimization:Artificial bee colony(ABC)algorithm. Journal of Global Optimization, 2007, 39:459-471.
[13] Karaboga D, Basturk B. On the performance of artificial bee colony(ABC)algorithm. Applied Soft Computation, 2008, 8:687-697.
[14] Karaboga D, Ozturk C. A novel clustering approach:Artificial Bee Colony(ABC)algorithm.Applied Soft Computation, 2011, 11:652-657.
[15] Feng J, Jiao L. Zhang X, et al. Hyperspectral band selection based on trivariate mutual information and clonal selection. IEEE Transactions on Geoscience Remote Sensing, 2014, 52(7):4092-4105.
[16] Feng J, Jiao L, Liu F, et al. Mutual-information-based semi-supervised hyperspectral band selection with high discrimination, high information, and low redundancy. IEEE Transactions on Geoscience Remote Sensing, 2015, 53(5):2956-2969.
[17] Mitra P, Murthy C A, Pal S K. Unsupervised feature selection using feature similarity. IEEE Transactions on Pattern Analysis Machine Intelligence, 2002, 24(3):301-312.
[18] Wu J. Unsupervised intrusion feature selection based on genetic algorithm and FCM. Lecture Notes in Electrical Engineering, 2012, 154:1005-1012.
[19] Martinez-Usómartinez-Uso A, Pla F, Sotoca J M, et al. Clustering-based hyperspectral band selection using information measures. IEEE Transactions on Geoscience Remote Sensing, 2007,45(12):4158-4171.
[20] Bazi Y, Melgani F. Toward an optimal SVM classification system for hyperspectral remote sensing images. IEEE Transactions on Geoscience Remote Sensing, 2006, 44(11):3374-3385.