论文部分内容阅读
BP神经网络具有收敛速度快和自学习、自适应功能强的特点,能最大限度地利用样本集的先验知识,自动提取合理的模型。本文采用Landsat TM遥感图像作为数据源,以山西省定襄县为研究区,通过主成分分析方法来压缩输入数据,并结合NDVI和纹理特征来建立BP神经网络的土地利用分类模型,将分类结果与基于光谱单元信息的神经网络分类和基于纹理特征的神经网络分类结果进行定性和定量比较分析。结果表明:该方法总精度达到了80.50%,分别比基于光谱单元信息的神经网络分类和基于纹理特征的神经网络分类提高了18.89%和6.23%,能够有效地解决地物光谱混淆、分类精度不高等问题。
BP neural network has the characteristics of fast convergence, self-learning and self-adaptive function, and can make use of the prior knowledge of the sample set to the maximum to automatically extract a reasonable model. In this paper, we use Landsat TM remote sensing image as the data source and Dingxiang County in Shanxi Province as the research area, compress the input data by principal component analysis and combine the NDVI and texture features to establish the land use classification model of BP neural network, With qualitative and quantitative comparative analysis of neural network classification based on spectral element information and texture classification based on neural network. The results show that the proposed method achieves an overall accuracy of 80.50%, which is 18.89% and 6.23% higher than the neural network classification based on spectral element information and the texture classification based on texture feature respectively, which can effectively resolve the spectrum confusion of objects and the classification accuracy is not good Higher problems.