论文部分内容阅读
BP神经网络算法结构简单、快捷,与传统的分类方法相比,分类精度较高,能较好地反映土地覆盖的分类结果。通过多年实地景观照片库结合土地利用图等地理辅助数据,采用BP神经网络算法对于田绿洲1999年和2002年两期ETM遥感影像进行土地覆盖分类。结果表明:地理辅助数据参与下的BP神经网络用于土地覆盖分类研究可获得较好的分类结果,平均分类精度可达到90%以上。通过精度分析认为神经网络算法在遥感图像分析与处理技术中具有很大的应用潜力。
BP neural network algorithm is simple and fast in structure, and has higher classification accuracy compared with traditional classification methods, which can well reflect the classification results of land cover. By using multi-year field photos database combined with geo-assist data such as land use maps, this paper uses BP neural network algorithm to classify land cover of ETM remote sensing images of Yutian Oasis in 1999 and 2002. The results show that the BP neural network with the help of geo-assistant data can get better classification results for land cover classification, and the average classification accuracy can reach more than 90%. Through the precision analysis, the neural network algorithm has great potential in remote sensing image analysis and processing technology.