It trains the SOM in a supervised way firstly, and then decomposes the mixed pixels based on fuzzy model.

• 首先对自组织映射神经网络进行有监督的训练，然后基于模糊模型对混合像元进行分解。