Abstract：Based on the radial basis function neural network (RBFNN) theory and the specialty of hyperspectral remote sensing data, the effective feature extraction model was designed, and those extracted features were connected to the input layer of RBFNN, finally the classifier based on radial basis function neural network was constructed. The hyperspectral image with 64 bands of OMIS Ⅱ made by Chinese was experimented, and the case study area was zhongguancun in Beijing. Minimum noise fraction (MNF) was conducted, and the former 20 components were extracted for further processing. The original data (20 dimension) of extraction by MNF, the texture transformation data (20 dimension) extracted from the former 20 components after MNF, and the principal component analysis data (20 dimension) of extraction were combined to 60 dimension. For classification by RBFNN, the sizes of training samples were less than 6.13% of the whole image. That classifier has a simple structure and fast convergence capacity, and can be easily trained. The classification precision of radial basis function neural network classifier is up to 69.27% in contrast with the 51.20% of back propagation neural network (BPNN) and 40.88% of traditional minimum distance classification (MDC), so RBFNN classifier performs better than the other three classifiers. It proves that RBFNN is of validity in hyperspectral remote sensing classification.
Key words：Hyperspectral remote sensing image;Radial basis function neural network(RBFNN);Classification
[1] FENG Li-hua(冯利华). Systems Engineering—Theory & Practice(系统工程理论与实践), 2003, 23(7): 137. [2] Hung C C, Kim Y, Coleman T L. A Comparative Study of Radial Basis Function Neural Networks and Wavelet Neural Networks in Classification of Remotely Sensed Data, World Automation Congress, 2002. Proceedings of the 5th Biannual, 2002, 13: 9. [3] Carpenter Gail A, Gjaja Marin N. IEEE Transactions on Geoscience and Remote Sensing, 1997, 35(2): 308. [4] Vassilas N, Charou E. Proceedings of the 1997 13th International Conference on Digital Signal Processing, DSP. Part 2, 1997. 995. [5] JIA Yong-hong, ZHANG Chun-sen, WANG Ai-ping(贾永红, 张春森, 王爱平). Journal of Xi’an University of Science and Technology(西安科技学院学报), 2001, 21(1): 58. [6] ZHAO Mei-fang, LUO A-li, WU Fu-chao, et al(赵梅芳, 罗阿理, 吴福朝, 等). Spectroscopy and Spectral Analysis(光谱学与光谱分析), 2006, 26(2): 377. [7] XIONG Yu-hong, WEN Zhi-yu, WANG Ming-yan, et al(熊宇虹, 温志渝, 王命延, 等). Spectroscopy and Spectral Analysis(光谱学与光谱分析), 2007, 27(1): 139. [8] LI Chun-min, LI Ke-rong, WANG Yun-hai(李春民, 李克荣, 王云海). China Mining Magazine(中国矿业), 2006, 15(9): 25. [9] LUO Jing, XIU Chun-bo(罗 菁, 修春波). Micro-Computer Information(微计算机信息), 2005, 21: 191. [10] LI Hong-yi, SHI Zhou, SHA Jin-ming, et al(李洪义, 史 舟, 沙晋明, 等). Chinese Journal of Applied Ecology(应用生态学报), 2006, 17(8): 1475. [11] ZHAO Xin-yu, FEI Liang-jun, FANG Shu-xing(赵新宇, 费良军, 方树星). Journal of Hydraulic Engineering(水利学报), 2006, 37(6): 717. [12] YAN Ping-fan, ZHANG Chang-shui(阎平凡, 张长水). Artificial Neural Networks and Simulated Evolution Computer(人工神经网络与模拟进化计算). Beijing: Tsinghua University Press(北京: 清华大学出版社), 2005. [13] MAO Jian-xu, WANG Yao-nan(毛建旭, 王耀南). Journal of System Simulation(系统仿真学报), 2001, 13(11): 146. [14] LUO Jian-cheng, ZHOU Cheng-hu(骆剑承，周成虎). Journal of Image and Graphics(中国图象图形学报), 2002, A7: 94. [15] XIONG Zhen, TONG Qing-xi, ZHENG Lan-fen(熊 桢, 童庆禧, 郑兰芬). Journal of Image and Graphics(中国图象图形学报), 2000, 5(3): 196. [16] Boardman J W, Kruse F A. Automated Spectral Analysis: A Geological Example Using AVIRIS Data-North Grapevine Mountains. Nevada: In Proceeding, ERIM Tenth Thematic Conference on Geologic Remote Sensing, Environmental Research Institute of Michigan, Ann Arbor, MI, I-407. [17] ZHANG Jin-shui, HE Chun-yang, PAN Yao-zhong, et al(张锦水, 何春阳，潘耀忠, 等). Journal of Remote Sensing(遥感学报), 2006, 10(1): 49. [18] Qian Du, Szu H. Lagrange Constrained Neural Network-Based Approach to Hyperspectral Remote Sensing Image Classification. Proceedings of 2003 International Conference on Neural Networks and Signal Processing, 2003. 270.
