基于神经网络和数据融合的红树林群落分类研究

Classification of Mangroves by Data Fusion and Neural Networks

及时准确地掌握红树林群落现状信息可为保护和修复红树林生态系统提供重要的决策依据。对红树林群落进行遥感分类在实际应用中具有较大的意义。但红树林各群落间的光谱差异很微弱,有必要采用多源遥感数据融合的方法来提高分类的精度。本文以珠海淇澳岛红树林区为例,使用SAR图像与TM图像,探讨了监督分类、非监督分类以及神经网络分类3种分类方法和IHS融合、小波融合以及主成分融合3种融合方法对红树林群落进行分类的效果。结果表明,对SAR与TM主成分融合图像应用神经网络分类方法能够取得最好的分类效果。

Guangdong in south China has the largest area of mangrove wetlands in the country. However, the mangrove wetlands have been rapidly diminishing because of the fast urbanization. Remote sensing can be used to obtain the information about mangrove wetlands, such as the monitoring of mangrove changes and classification of mangrove types. The information is very important for wetland conservation in rapid growing areas. It is very important to obtain the information about the changes of mangroves and the distribution of mangrove types, which is required for decision-making. In this study, SAR and TM images are used to classify the mangrove using unsupervised, supervised, neural networks and data fusion methods. And three types of data fusion such as IHS, Wavelet and PCA are used in mangrove classification. The study demonstrates that radar remote sensing can provide useful information for mangrove study. It is able to produce more accurate classification because of its excellent capability of penetration than optical remote sensing. This study indicates that the PCA fusion of SAR images with TM images can provide better classification of mangrove than other common methods.