融合空间信息的基于稀疏约束的非负矩阵分解的高光谱图像自适应端元提取
PTIVE ENDMEMBER EXTRACTION BASED SPARSE NONNEGATIVE MATRIX FACTORIZATION
本文针对高光谱图像处理中端元个数难得精确估计和端元的可变性的问题，提出了一种融合空间信息的基于稀疏约束的非负矩阵分解的自动端元提取算法。该自适应端元提取算法利用超像素分割，和基于稀疏约束的非负矩阵分解框架，考虑了高光谱图像的局部和非局部信息，针对端元光谱的可变性，能够自适应地提取端元。实验验证了算法的有效性，为后续的混合像元分解提供了一种解决思路。
In this paper，an adaptive endmember extraction base sparse nonnegative matrix fatorization method was proposed. With superpixel segmentation, the proposed method is based on sparse nonnegative matrix factorization.framework. considering the endmembers' varibility, the proposed method show better performance than the classical endmember extraction method, which provides a new way of spectral unmixing.. (10 Points, Times New Roman)
陈香香、李登刚、李华丽
遥感技术
高光谱图像处理空间信息端元提取
Hyperspectral image processingspatial informationendmember extraction
陈香香,李登刚,李华丽.融合空间信息的基于稀疏约束的非负矩阵分解的高光谱图像自适应端元提取[EB/OL].(2017-05-16)[2025-08-16].http://www.paper.edu.cn/releasepaper/content/201705-922.点此复制
评论