However,the endmember spectral variability commonly exists,which leads to the lower accuracy of pixel unmixing.
传统方法假设每个端元具有稳定的光谱特征,然而端元内光谱差异普遍存在,这将导致混合像元分解精度的降低。
In this paper, by use of actual most different spectra to construct original vectors, a new target transformation factor analysis was developed.
提出一种新的目标转化因子分析，其初始向量的构造基于实际最大差异光谱，无需分离即可从红外混合体系中解析出纯组份光谱。