基于相容粗糙集的BP神经网络湿地覆被信息提

发布时间：2018-04-17 11:30

本文选题：相容粗糙集 + BP神经网络　；参考：《资源科学》2016年08期

【摘要】：BP神经网络因具有自学习、自适应、大规模并行处理等特点而广泛应用于遥感影像分类中,但是该方法训练时容易陷入局部极小值,且收敛速度较慢,针对这些不足提出一种基于相容粗糙集的BP神经网络分类方法。本文以双台子河口湿地为研究对象,以Landsat-8 OLI影像为数据基础,利用相容粗糙集理论对样本数据集进行预处理,将得到的数据作为新的训练样本,在Matlab软件平台下建立BP神经网络的湿地覆被分类模型,进行湿地覆被信息提取,将分类结果与单纯的BP神经网络以及粗糙集样本属性约简预处理的分类结果进行比较分析。结果表明,基于相容粗糙集的BP神经网络分类方法可以剔除训练样本中的噪声数据,提高网络的训练成功率,缩短网络的收敛时间,分类效果较好,其总体精度达到91.25%,Kappa系数为0.8969,比单纯的BP神经网络分类结果高7.92%和0.0926,比粗糙集样本属性约简预处理方法的分类结果高3.03%和0.0357,是一种有效的湿地覆被分类方法。
[Abstract]:BP neural network is widely used in remote sensing image classification because of its characteristics of self-learning, self-adaptation, large-scale parallel processing and so on.A BP neural network classification method based on compatible rough set is proposed to solve these problems.Taking Shuangtaizi Estuary Wetland as the research object and Landsat-8 OLI image as the data base, using the consistent rough set theory to preprocess the sample data set, the obtained data is taken as a new training sample.The wetland cover classification model based on BP neural network was established based on Matlab software, and the wetland cover information was extracted. The classification results were compared with those of BP neural network and rough set sample attribute reduction preprocessing.The results show that the classification method of BP neural network based on consistent rough set can eliminate the noise data from the training samples, improve the training success rate of the network, shorten the convergence time of the network, and the classification effect is better.The total accuracy of the method is 0.8969, which is 7.92% and 0.0926 higher than that of BP neural network alone, and 3.03% and 0.0357 higher than that of rough set sample reduction preprocessing method. It is an effective wetland cover classification method.
【作者单位】：沈阳农业大学水利学院;
【基金】：国家自然科学青年科学基金项目(31200392)
【分类号】：P237

【相似文献】