摘要
根据总体最小二乘(total least squares,TLS)模型理论,提出了一种影像端元光谱可受噪声污染的混合光谱线性扩展模型,并实现了该模型的端元光谱自动迭代提取以及混合像元的限定性分解。实验结果表明,扩展的混合像元分解模型明显优于传统的最小二乘分解模型,总体精度大约提高了10%~20%。
An extension linear spectral model based on total least square algorithm is proposed, and iteratively abstract the better endmembers and physically based constrained algorithm with this model are presented. The results show that the proposed model provides better accuracy than tradi- tional least square method, the total estimating precision improves by 10%-20%.
出处
《武汉大学学报(信息科学版)》
EI
CSCD
北大核心
2008年第5期457-460,468,共5页
Geomatics and Information Science of Wuhan University
基金
国家科技部国际科技合作重点资助项目(2005DFA20420)
福州大学引进人才启动基金资助项目
关键词
总体最小二乘
混合像元
端元光谱
迭代分解
Abstracting
Boolean functions
Curve fitting
Geodesy
Least squares approximations