结合RBF网络与光谱指数的遥感分类应用研究.pdfVIP

Hans Journal of Wireless Communications 无线通信, 2014, 4, 136-142 Published Online December 2014 in Hans. /journal/hjwc /10.12677/hjwc.2014.46021 Application Study on Remote Sensing Classification Based on RBF Network and Spectral Indexes 1 1 1,2 Sanjun Huang , Yingying Hao , Xiaobo Luo 1 College of Computer, Chongqing University of Post Technical College, Chongqing 2 College of Resources and Environment, Southwest University, Chongqing Email: luoxb@ rd th th Received: Oct. 23 , 2014; revised: Nov. 17 , 2014; accepted: Nov. 26 , 2014 Copyright © 2014 by authors and Hans Publishers Inc. This work is licensed under the Creative Commons Attribution International License (CC BY). /licenses/by/4.0/ Abstract In this paper, spectral indexes NDWI, NDVI, NDBI were inversed from TM images as the key aux- iliary information in the classification of city land use. On this basis, the city remote sensing classi- fication model was put forward based on RBF network and the normalized indexes. Finally, taking the Sichuan Nanchong city as the study area, using TM image as data source, the city classification model proposed in this paper was experimented. The experimental results show that RBF net- work has a certain advantage in the integration of learning parameters. The overall accuracy us- ing RBF neural network and the surface indexes can reach to 95.02%, which is improved by 7.05 percentage points than only using the band information. Keywords RBF Neural Network, Spectral Index, Urban Land Use, Classification 结合RBF网络与光谱指数的遥感分类应用研究 1 1 1,2 黄三军 ，郝莹莹，罗小波 1重庆邮电大学计算机学院，重庆