植物叶片生化组分遥感反演的主成分变换算法

Remote Sensing Inversion of Leaf Biochemical Components through PCA Algorithm

主成分分析(PCA)算法是一种常见的高光谱数据特征提取方法。针对PROSPECT辐射传输模型反演问题,尝试了两种PCA算法来对高光谱数据进行变换,进而反演植被生化组分含量。反演结果表明:两种PCA反演算法均能对传统反演算法中干物质难反演的问题有所改善;分块主成分算法比全局主成分算法具有更好的反演效果。

PCA is an often used algorithm to extract the feature information from hyperspectral data. For the inver- sion of radiative transfer model PROSPECT, two kinds of PCA algorithm was used to transform the hyperspectral data, then the feature information was applied to retrieve the biochemical components of vegetation. The results show that both two kinds of PCA algorithm can remarkably improve the inversion precision of dry matter compared with traditional inversion method and the block PCA shows better effects than the global PCA.