散射项约束非负矩阵分解的高光谱图像解混

针对传统的约束非负矩阵分解方法对于解混的物理特性考虑较少,提出一种高光谱图像的解混方法:散射项约束非负矩阵分解(scattering-term constrained nonnegative matrix factorization,STC-NMF).与大多数约束非负矩阵分解算法将约束建立在数据的数学特性之上不同,STC-NMF考虑到大气中的悬浮物、胶着物的散射作用,对成像光谱仪接收到的光谱信号有着不可忽视的影响,将大气中的米氏散射(Mie scattering)造成的邻域贡献视作干扰,通过将散射相函数作为约束条件的约束非负矩阵分解,在丰度上对目标像素及其邻域的米氏散射干扰进行约束,以达到在目标函数上将米氏散射和噪声造成的干扰降低到最小的效果,通过实验验证了提出方法的有效性.

Scattering-term constrained nonnegative matrix factorization for hyperspectral unmixing

In view of the fact that the traditional constrained nonnegative matrix factorization seldom considers the physical characteristics of the unmixed objects, this paper proposes a method for hyperspectral unmixing:scattering-term constrained nonnegative matrix factorization (STC-NMF). Different from the most constrained nonnegative matrix factorization, STC-NMF considers the scattering of suspended matter and adhesive matter in atmosphere which has a nonnegligible impact on the spectral signals received by the imaging spectrometer. The neighborhood contribution caused by Mie scattering in the atmosphere is regarded as interference, so the Scattering-phase Function is regarded as constraint non-negative matrix factorization. And this constraint constrains the interference of Mie Scattering of the target pixel and its neighborhood in terms of abundance, in order to minimize the interference caused by Mie scattering and noise in the objective function, and the effectiveness of the proposed method is verified by experiments.