神经网络技术在成像光谱数据的岩石矿物识别中的应用

由于遥感图象的复杂性，在遥感图像上直接进行岩石矿物识别是比较困难的，成像光谱数据的应用和人工神经网络模拟技术的日益成熟，为利用遥感图像直接进行岩石矿物识别提供了一种非常有效的方法。该文利用简化的二阶B－P神经网络，用标准的矿物波谱曲线作为训练样本，对试验区内的OH－类矿物进行识别试验，取得了很好的效果。试验表明：利用神经网络技术进行成像光谱数据的岩石矿物识别是非常有效的。

Recognise Lithology and Mineral in Remote Sensing Imagery Using Neural Network

It is difficult to recognise directly lithology and mineral in remote sensing imagery. With progress in application of imaging spectrum data and neural network, methords recognising lithology and mineral using remote sensing imagery are already very effective. This paper study high order B-P neural network. Using standard mineral spectrum data as training sample, recognise hydroxyl-bearing mineral in testing area. As a result, it is effective to recognise lithology and mineral in remote sensing imagery using high order neural network.