通过对比分析MNDWI结合阈值法和波段谱间关系结合阈值法这两种常见的水体信息法,针对其识别山体阴影和水体的准确率的不高的缺点,提出了一种新的光谱变换结合神经网络分类法。实验结果表明,新光谱变换结合BP神经网络的遥感分类方法基本上消除了山体阴影对水体提取的干扰,提取水体信息的效果比前两种方法要好。 By comparing and analyzing MNDWI combined threshold method with band spectrum relationship and threshold method, two common water body information methods are proposed. In view of their shortcomings of recognizing the mountain shadow and the low accuracy of water bodies, a new spectral transformation combined with nerve Network taxonomy. Experimental results show that the new spectral transform combined with BP neural network remote sensing classification method basically eliminates the interference of mountain shading on water extraction, and the effect of extracting water body information is better than the former two methods.