摘要
像元组分分解模型主要包括线性混合模型(LSMM)、模糊分类器模型、神经网络模型以及高斯混合判别分析模型(MDA)。以线性光谱混合模型为例给出了像元组分分解的过程,并分析了各种模型在不同领域的应用。像元组分分解还处于探索阶段,应根据不同的应用领域,不同的前提条件来选择和发展更为合适的模型,或用几种模型进行组合。端元的类型和数量的确定要力求准确,端元光谱值的获取要根据实际分析,研究更为适合的方法,使组分像元分解的精度得到进一步的提高。
Models of components unmixing of mixed pixel include mainly the line spectral mixture model （LSMM）, fuzzy classifier model, neural network model, and Gaussian mixture discriminant analysis （MDA） model. The process of pixel unmixing is presented as an example from LSMM. Then, some application examples are analyzed by the pixel unmixing techniques. Since the components unmixing of mixed pixel presently is just in an exploring phase, the different models or the combination of several models should be chosen based on different application areas and different premises. The class and number of endmember should be exact. The spectral value of endmember needs to be analyzed and considered strictly based on some technique to make sure the precision excellent.
出处
《世界地质》
CAS
CSCD
2006年第2期201-207,共7页
World Geology
基金
国家自然科学基金项目(40471086)资助
关键词
遥感
混合像元
像元组分分解
模型
端元
remote sensing
mixed pixel
pixel components unmixing
model
endmember
