基于单体扩张的端元提取算法

Endmember Extraction Algorithm Based on the Simplex Expansion

端元提取是高光谱混合像元分解的重要环节。为了提取高光谱图像的端元,本文基于线性表示理论与凸锥模型理论,论证了:与单体共面的单体外向量被单体的顶点向量线性表示时,表示系数必有负值,从而给出了理想情形下判别端元的充要条件,并在此基础上,针对非理想情形提出了一种提取端元的迭代算法。实验结果表明,算法提取端元的精度优于VCA算法、效率高于搜索算法,算法稳定性好,对噪声的敏感性低。

Endmember extraction is a key step in unmixing hyperspectral mixed pixels. In order to extract endmembers of hyperspectral image, this paper proves that if a vector in vitro and vivo is represented by a vertex vector, there will be a negative coefficient based on the theory of linear representation and the theory of convex cone model. The theory gives a necessary and sufficient condition for the endmembers identification. A new iterative algorithm is proposed under non-ideal situation. The experimental results show that the precision of the endmember extraction using the algorithm proposed in this paper is better than VCA algorithm. And this algorithm has better efficiency than the search algorithm. It has good stability and a low sensitivity to noise.