摘要
传统遥感影像线性拟合方法实际效果较不理想。为此,提出一种基于BP神经网络的方法进行遥感影像波段模拟,应用其良好的非线性拟合能力实现对影像的光谱特征处理。以ETM传感器的通道2—4作为已知样本,通道1蓝光波段作为目标向量进行大量BP网络训练,得到优化网络模型后再进行精度验证,最后将模型用于拟合SPOT影像蓝光波段。经验证,该方法可在一定程度上解决ETM和SPOT多光谱影像波段缺失的问题,有一定使用价值。
Traditional method of remote sensing images simulation is not so good.Therefore,a method based on BP neural network in band simulation is proposed,which can processspectral signature of images under the ability of non-linear fitting.ETM channel 2—4 are used as sample data.Channel 1,the blue band is used as target vector according to the BP networks regulation.After the sample training,the optimized network model is obtained,which precision will be verified.Finally,the model can be used to simulate the blue band of SPOT image.It is proved that the problem of band missing can be solved to a certain extent by using the method in this study.
出处
《现代测绘》
2018年第1期44-46,共3页
Modern Surveying and Mapping
关键词
波段模拟
BP神经网络
精度评定
band simulation
BP neural network
precision assessment
