论文部分内容阅读
高光谱图像中,单一端元光谱很难准确刻画一个类别,导致解混结果不准确。针对经典多端元光谱解混(MESMA)算法存在计算量大、端元预选繁琐等缺点,提出基于分层的MESMA(HMESMA)算法,第1层确定像元包含地物类别,第2层在第1层的基础上再分层确定像元包含最优端元个数。采用模拟数据和真实高光谱数据进行实验,证明了本文算法比固定端元解混效果好,平均丰度误差最高降低了2.65%,与经典的MESMA算法精度相当,但大大降低了计算量,提高了计算效率。
Hyperspectral images, a single end-spectrum is difficult to accurately characterize a category, resulting in inaccurate mixing results. Aiming at the shortcomings of large computational complexity and complicated preselection of endmember, the MESMA algorithm based on stratification is proposed. The algorithm of HMESMA is proposed in the first step, Based on the layer 1, the cells are further delaminated to determine the number of pixels containing the optimal endmember. Experiments with simulated data and real hyperspectral data demonstrate that the proposed algorithm has better de-mixing effect than fixed-end elements, with a mean maximum error of 2.65%, which is equivalent to that of the classical MESMA algorithm. However, the computational complexity is greatly reduced The calculation of efficiency.