基于Cayley-Menger行列式的高光谱遥感图像端元提取方法

Cayley-Menger determinant-based endmember extraction algorithm for hyperspectral unmixing

提出了一种基于Cayley-Menger行列式的快速端元提取算法.该算法的目标是寻找包含高光谱数据集的最小体积的单形体.与其它基于单形体几何的算法相比,该方法具有诸多优点.首先,Cayley-Manger行列式的引入使得算法可以便捷地利用Hermite矩阵的特点大大加速搜索过程,进而得到一个稳定的最终解.其次,该算法无须对数据进行降维处理,从而可以避免因数据降维而造成的有用信息的丢失.仿真和实际高光谱数据的实验结果表明,所提出的算法在获得准确解的同时,具有非常快的收敛速度.

A fast Cayley-Menger determinant-based endmember extraction algorithm for hyperspectral unmixing was proposed. The algorithm is to find the simplex enclosing the hyperspectral data with minimum volume. It improves current simplex-based algorithms in several aspects. The introduction of Cayley-Menger determinant makes it easy to use features of Hermite matrix to accelerate the searching process and gives a stable result finally. Moreover, a dimensionality reduction transform is not necessary in this algorithm, which will avoid the loss of useful information during the dimensionality reduction. The experimental results on synthetic and real hyperspectral dataset demonstrated that the proposed algorithm is a fast and accurate algorithm for the hyperspectral unmixing.