基于贝叶斯非负矩阵分解的区域关联型像元分解

Area-correlated spectral unmixing based on Bayesian nonnegative matrix factorization

研究了区域关联型像元分解技术,旨在建立相邻或相似区域、不同高光谱图像之间的空间联系性,提高像元分解结果的可比性与综合解析能力。针对独立型像元分解技术存在的问题,提出了一种基于贝叶斯非负矩阵分解的区域关联型像元分解算法,根据相邻区域图像的像元分解结果估计当前图像各个端元及其丰度值的先验概率密度函数,并通过采样的方法进行像元分解,估计端元矩阵和丰度值矩阵。模拟数据的定量评价结果表明该算法与实际值具有更好的相似度,真实数据的定量分析结果则验证了该算法的实际有效性。

A study of area-correlated spectral unmixing was conducted to establish the spatial correlation between different hyperspectral images in adjacent or similar areas to improve the comparable analysis for spectral unmixing.To solve the problem of the spatial correlation in adjacent areas for traditional spectral unmixing methods,an area-correlated spectral unmixing method based on Bayesian nonnegative matrix factorization was proposed.When using the proposed method,the spatial correlation property between two adjacent areas is expressed by a priori probability density function,and the endmembers extracted from one of the adjacent areas are used to estimate the priori probability density functions of the endmembers in the current area,which works as a type of constraint in the iterative spectral unmixing process.The quantitative evaluation on synthetic hyperspectral images indicates that the results of the proposed method are more similar to the real values,while the quantitative analysis on real hyperspectral images demonstrates the effectivity of the proposed method.