鄱阳湖区湿地特征提取研究

发布时间：2018-03-23 23:37

本文选题：鄱阳湖　切入点：湿地　出处：《江西理工大学》2014年硕士论文

【摘要】：随着遥感影像应用越来越广泛，利用遥感影像提取湿地特征，对湿地信息进行科学的获取与管理，为湿地和湿地物种的保护提供了一种新途径。而鄱阳湖湿地是亚洲面积最大的湿地，资源丰富，类型众多，极具代表性。因此，，利用遥感技术对鄱阳湖区湿地特征进行分类提取研究具有现实意义。本文首先介绍了遥感影像分类以及湿地信息提取方面的国内外研究现状；其次，介绍了鄱阳湖区概况，并对遥感影像数据进行了预处理，采集了样本数据；接着，采用ISODATA分类法、K-Means分类法、平行六面体分类法、最小距离法、最大似然法、决策树算法和BP神经网络算法分别对鄱阳湖区都昌县湿地特征进行分类提取，得到分类结果并评价其分类效果；然后，对决策树算法和BP神经网络算法进行了集成研究，提出一种基于决策树与BP神经网络的集成算法，并利用该算法实现了都昌县湿地特征分类提取，得到了分类结果并进行了分类精度评价；最后，对文中运用的分类方法的分类精度进行了综合对比分析，并采用集成算法实现了鄱阳湖区湿地特征提取。通过实验得到，ISODATA分类法、K-Means分类法、平行六面体分类法、最小距离法、最大似然法、决策树算法、BP神经网络算法以及集成算法的总体分类精度分别为：75.58%、82.16%、75.46%、77.63%、82.69%、85.03%、89.96%和92.93%。实验结果表明，基于决策树与BP神经网络的集成算法在湿地特征提取中分类精度明显高于其它算法，从而为湿地信息的获取提供了一种新方法。
[Abstract]:With the application of remote sensing image more and more widely, the characteristics of wetland are extracted from remote sensing image, and the information of wetland is obtained and managed scientifically. Poyang Lake wetland is the largest wetland in Asia, rich in resources, numerous types and representative. It is of practical significance to study the classification and extraction of wetland features in Poyang Lake area by remote sensing technology. This paper first introduces the classification of remote sensing images and wetland information extraction at home and abroad research status; secondly, introduced the Poyang Lake region, and the remote sensing image data preprocessing, collected the sample data; ISODATA classification, parallel hexahedron classification, minimum distance method, maximum likelihood method, decision tree algorithm and BP neural network algorithm were used to classify and extract the wetland features of Duchang County in Poyang Lake region. Then, the decision tree algorithm and BP neural network algorithm are integrated, and an integration algorithm based on decision tree and BP neural network is proposed. The algorithm is used to extract the wetland features of Duchang County, and the classification results are obtained and the classification accuracy is evaluated. Finally, the classification accuracy of the classification method used in this paper is compared and analyzed. The integrated algorithm is used to extract the wetland features in Poyang Lake region. The total classification accuracy of K-Means classification, parallel hexahedron classification, minimum distance method, maximum likelihood method, decision tree algorithm and BP neural network algorithm are 85.58 / 75.4677.630.89. 96% and 92.933% respectively. The experimental results show that. The classification accuracy of the integrated algorithm based on decision tree and BP neural network is obviously higher than that of other algorithms in wetland feature extraction, which provides a new method for wetland information acquisition.
【学位授予单位】：江西理工大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：P237

【参考文献】