A New Method of Fuzzy Supervised Classification of High Resolution Remote Sensing Image
-
摘要: 针对高分辨率遥感影像分类中由于细节特征突出、同质区域光谱测度变异性增大所带来的像素类属的不确定性及模型的不确定性等造成的误分结果，提出一种基于模糊隶属函数的监督分类方法。对同质区域定义高斯隶属函数模型用来表征像素类属不确定性；模糊化该隶属函数参数建立影像模糊隶属函数，以建模同质区域光谱测度的不确定性；用训练样本在所有类别中的模糊隶属函数及原隶属函数（高斯隶属函数）中的隶属度为输入，建立模糊线性神经网络模型作为目标函数，实现分类决策。该算法和经典算法对World View-2全色合成影像及真实影像进行定性和定量分类实验，分类结果验证了文中方法具有更高的分类精度。Abstract: This paper presents a supervised image classification method based on fuzzy membership function to solve incorrect classification of high resolution remote sensing image, which caused by highlight detail information, the uncertainly of the pixels classification derived from the increase of the differences between pixels in the homogenous region, the uncertainly of classification decision and so on. First, Gaussian model is used to characterize the uncertainly of the membership of pixels; then the model is extended to build the image fuzzy membership function to define the uncertainly of the homogenous regions. To segment the image, the objective function is built by linear function of neural network, which the fuzzy membership functions and the membership degrees of the original fuzzy membership functions as input values. The proposed method is compared with the classification methods tested on the WorldView-2 panchromatic synthetic and real images. Through the qualitative and quantitative experiments, it can be found that the proposed method has better classification accuracy.
-
-
![]()
图 5 高分辨率遥感影像及分类结果
Figure 5. High Resolution Remote Sensing Images and the Classification Results
表 1 合成影像分类结果的用户精度、产品精度、总精度及kappa值
Table 1 User Precision, Product Precision, Total Precision and Kappa Value of Synthetic Image
方法 精度指标 同质区域 调节因子(b/a) 总精度 kappa值 Ⅰ Ⅱ Ⅲ Ⅳ 高斯隶属函数 用户精度 0.765 0.893 1.000 0.781 0.856 0.809 产品精度 0.786 1.000 0.882 0.758 最大似然 用户精度 0.816 0.925 1.000 0.897 0.906 0.874 产品精度 0.907 1.000 0.921 0.795 模糊化均值 用户精度 0.819 0.988 0.998 0.927 c1= c2= c3= c4= 2.5 0.930 0.906 产品精度 0.938 0.997 0.988 0.795 模糊化标准差 用户精度 0.820 0.988 0.998 0.927 d1= d2= d3= d4= 0.3 0.930 0.906 产品精度 0.938 0.998 0.988 0.795 
下载: 导出CSV
表 2 各类别训练数据分类结果的用户精度、产品精度、总精度及kappa值
Table 2 User Precision, Product Precision, Total Precision and Kappa Value of High Resolution Images
影像 地物 隶属函数法 最大似然法 本文方法 影像 用户
精度产品
精度总精度
/kappa用户
精度产品
精度总精度
/kappa用户
精度产品
精度总精度
/kappa图 5(a) 道路 1.000 0.980 0.989/0.980 1.000 0.980 0.989/0.980 0.998 1.000 0.998/0.996 植被 0.976 0.999 0.976 0.999 1.000 9.995 水域 0.994 0.992 0.994 0.992 0.973 0.998 图 5(b) 建筑物 0.901 0.636 0.907/0.881 0.890 0.940 0.952/0.937 0.942 0.970 0.991/0.988 植被 0.961 0.964 0.965 0.959 0.971 0.982 积雪 0.551 0.843 0.845 0.752 0.990 0.991 水域 0.998 1.000 0.998 0.999 0.999 1.000 冰面 0.978 0.981 0.974 0.986 0.997 0.986 图 5(c) 建筑物 0.919 0.786 0.904/0.849 0.836 0.829 0.914/0.865 0.931 0.793 0.932/0.892 植被 0.772 0.938 0.862 0.941 0.860 0.947 水域 0.980 0.948 0.990 0.944 0.992 0.996
下载: 导出CSV
-
[1] Bruzzone L, Carlin L. A Multilevel Context-based System for Classification of Very High Spatial Re-solution Images[J]. IEEE Transactions on Geoscience and Remote Sensing, 2006, 44(9):2587-2600 https://www.deepdyve.com/lp/elsevier/qualitative-satellite-image-analysis-mapping-spatial-distribution-of-LPZkqE0g2x
[2] Gath I, Geva A B. Unsupervised Optimal Fuzzy Clustering[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 1989, 11(7):773-780 http://cn.bing.com/academic/profile?id=6980625f9c7bd2be5e88ee9100f0b0a8&encoded=0&v=paper_preview&mkt=zh-cn
[3] Ahmed M N, Yamany S M, Mohamed N A, et al. A Modified Fuzzy C-means Algorithm for Bias Field Estimation and Segmentation of MRI Data[J]. IEEE Transactions on Medical Imaging, 2002, 21(3):193-199 doi: 10.1109/42.996338
[4] Chen S, Zhang D. Robust Image Segmentation Using FCM with Spatial Constraints Based on New Kernel-induced Distance Measure[J]. IEEE Tran-sactions on Cybernetics Systems, Man and Cyberne-tics, 2004, 34(4):1907-1916 doi: 10.1109/TSMCB.2004.831165
[5] Szilagyi L, Benyo Z, Szilagyi S M, et al. MR Brain Image Segmentation Using an Enhanced Fuzzy C-means Algorithm[C]. Proceedings of the 25th Annual International Conference of the IEEE Enginee-ring in Medicine and Biology Society, Cancun, Mexico, 2003
[6] 王海军, 邓羽, 王丽, 等.基于数据场的C均值聚类方法研究[J].武汉大学学报·信息科学版, 2009, 34(5):626-629 http://ch.whu.edu.cn/CN/abstract/abstract1268.shtml Wang Haijun, Deng Yu, Wang Li, et al. A C-means Algorithm Based on Data Field[J]. Geoma-tics and Information Science of Wuhan University, 2009, 34(5):626-629 http://ch.whu.edu.cn/CN/abstract/abstract1268.shtml
[7] Chatzisand S P, Varvarigon T A. A Fuzzy Clustering Approach Toward Hidden Markov Random Field Models for Enhanced Spatially Constrained Image Segmentation[J]. IEEE Transactions on Fuzzy System, 2008, 16(5):1351-1361 doi: 10.1109/TFUZZ.2008.2005008
[8] Sarkar A, Banerjee A, Banerjee A, et al. Land Co-ver Classification in MRF Context Using Dempster-Shafer Fusion for Multisensory Imagery[J]. IEEE Transactions on Image Processing, 2005, 14(5):634-645 doi: 10.1109/TIP.2005.846032
[9] Karmakar S, Mujumdar P P. Grey Fuzzy Optimization Model for Water Quality Management of a Rriver System[J]. Advances in Water Resources, 2006, 29(7):1088-1105 http://cn.bing.com/academic/profile?id=82b2d8ce3535479c2442637b2547eb20&encoded=0&v=paper_preview&mkt=zh-cn
[10] Oberthur T, Dobermann A, Aylward M. Using Auxiliary Information to Adjust Fuzzy Membership Functions for Improved Mapping of Soil Qualities[J]. Geographical Information Science, 2000, 14(5):431-454 doi: 10.1080/13658810050057588
[11] Mauro A E, Maros A E, Sinito A M. An Interval Fuzzy Model for Magnetic Monitoring:Estimation of a Pollution Index[J]. Environmental Earth Sciences, 2012, 66(5):1477-1485 doi: 10.1007/s12665-011-1387-z
[12] Janin S, Khare M. Construction of Fuzzy Membership Functions for Urban Vehicular Exhaust Emissions Modeling[J]. Environmental Monitoring and Assessment, 2010, 167(8):1-4 http://cn.bing.com/academic/profile?id=ed23b69292d44479d6265a332fa87b96&encoded=0&v=paper_preview&mkt=zh-cn
[13] Yang C L, Kuo R J, Chien C H, et al. Non-dominated Sorting Genetic Algorithm Using Fuzzy Membership Chromosome for Categorical Data Clustering[J]. Applied Soft Computing, 2015, 30(5):113-122 http://cn.bing.com/academic/profile?id=43bbb631e516c8875f5682d27bcf6089&encoded=0&v=paper_preview&mkt=zh-cn
[14] Tang W M. Fuzzy SVM with a New Fuzzy Membership Function to Solve the Two-Class Problems[J]. Neural Process Letters, 2011, 34(3):209-219 doi: 10.1007/s11063-011-9192-y
[15] Chen M S, Wang S W. Fuzzy Clustering Analysis for Optimizing Fuzzy Membership Functions[J]. Fuzzy Sets and Systems, 1999, 103(2):239-254 doi: 10.1016/S0165-0114(98)00224-3
[16] Kaur D, Mukherjee S, Basu K. Solution of a Multi-Objective and Multi-Index Real-Life Transportation Problem Using Different Fuzzy Membership Functions[J]. Journal of Optimization Theory and Applications, 2015, 164(2):666-678 doi: 10.1007/s10957-014-0579-6
[17] Kumar A, Kaur M. A New Method for Solving Single and Multi-Objective Fuzzy Minimum Cost Flow Problems with Different Membership Functions[J]. Indian Academy of Sciences, 2014, 39(1):189-206 http://cn.bing.com/academic/profile?id=9d13b8ce84af93f504f7d37a90911b68&encoded=0&v=paper_preview&mkt=zh-cn
[18] Mccloskey A, Mcivol R, Maguire L. A User-centred Corporate Acquisition System:A Dynamic Fuzzy Membership Functions Approach[J]. Decision Support Systems, 2006, 42(1):162-185 https://www.deepdyve.com/lp/elsevier/a-user-centred-corporate-acquisition-system-a-dynamic-fuzzy-membership-Zb21XSpy43
[19] Qiu F, Jensen J R. Opening the Black Box of Neural Networks for Remote Sensing Image Classification[J]. International Journal of Remote Sensing, 2004, 25(9):1749-1768 http://cn.bing.com/academic/profile?id=65297852c9fcc3de64101413e0538b94&encoded=0&v=paper_preview&mkt=zh-cn
[20] Liu J, Wang W, Dou X. Multiple Model Internal Model Control Based on Fuzzy Membership Function[C]. 2008 IEEE International Conference on Automation & Logistics, Qingdao, China, 2008
计量
- 文章访问数: 1518
- HTML全文浏览量: 165
- PDF下载量: 339
