NOAA卫星图像水体信息神经网络识别方法的探讨

资助项目: 本文由“湖北省极轨气象卫星遥感农业应用服务系统”项目资助

A DISCUSSION ON APPLICATION OF NEURAL NETWORK TECHNIQUE IN IDENTIFICATION OF WATER BODY FROM NOAA SATELLITE IMAGES

摘要: 该文阐述了应用神经网络识别NOAA卫星图像水体信息的基本原理和方法, 并进行了实例分析, 结果表明, 神经网络法比阈值法具有更高的精度和效率.
- 水体识别;
- 神经网络;
- BP模型;
- NOAA卫星图像
Abstract: The principles and methods for using artificial neural network techniques to automatically identify the information about water body from NOAA satellite images are discussed. An example of its application is presented. The results show that the precision and efficiency using the neural network technique is higher than using the threshold methods.
- Water body identification;
- Artificial neural network;
- BP model;
- NOAA satellite image

图 1 BP网络的拓扑结构

下载: 全尺寸图片幻灯片

图 2 经神经网络处理前后的1998年8月4日NOAA卫星图像

(a) 处理前3个通道 (1, 2, 4) 合成图 (b) 处理后水体分布图 (黑色为水体)

下载: 全尺寸图片幻灯片

表 1 训练样本的特性输出表

表 2 水体识别结果的混淆矩阵表

[1]	盛永伟, 肖乾广, 陈维英, 等.气象卫星大范围洪涝灾害快速宏观动态监测方法研究.气象卫星遥感技术为农业服务应用研讨会文集, 北京:气象出社版, 1996.51~62.
[2]	孙涵, 居为民, 汤志成.江苏省洪涝灾害气象卫星遥感监测, 气象卫星遥感技术为农业服务应用研讨会文集, 北京:气象出社版.1996.45~48.
[3]	Paola J D, Schowengerdt R A.A detailed comparison of backpropagation neural network and maximum likelihood classifiers for urban landuse classification.IEEE Transactions on Geoscience and Remote Sensing, 1995, 33 (4): 81~995. http://ieeexplore.ieee.org/document/406684/
[4]	Bernard A C, Wilkinson G G, Kanellopoulos I.Training strategies for neural network soft classification of remotely-sensed lmagery, Int.J.Remote Sensing, 1997, 18 (8): 1851~1856. doi: 10.1080/014311697218160
[5]	Benediktsson J A, Sveinsson J R.Feature extraction for multisource data classification with a rtifical neural networks.Int.J.Remote Sensing, 1997, 18 (4): 727~740. doi: 10.1080/014311697218728
[6]	李学桥, 马莉.神经网络·工程应用.重庆:重庆大学出版社, 1996.37~44.