基于高光谱特征提取的藻类叶绿素a反演模型

Chl-a Inversion Model Based on Hyperspectral Feature Extraction of Algaes

选取自然界中分布较广泛的小球藻和铜绿微囊藻为研究对象,进行室内藻类光谱试验,同步测定其ρ(叶绿素a),并分别建立基于混合高斯函数的小球藻和铜绿微囊藻高光谱信息模型.在此基础上,利用模拟退火算法实现模型的非线性参数拟合,提取这2种藻的高光谱特征,通过非线性回归分析,反演得到分解后的高光谱信息模型的峰高(hi)与ρ(叶绿素a)的定量模型,实现对水体中小球藻和铜绿微囊藻ρ(叶绿素a)的预测.结果表明:藻类高光谱特征提取算法能有效揭示小球藻和铜绿微囊藻的光谱本质特征,并得出相应的小球藻和铜绿微囊藻叶绿素a反演模型.

Chlorella vulgafis and Microeystis aeruginos, found widely in natural environments, were selected for an indoor experiment. The reflectance spectra of the algaes were measured and the chlorophyU-a concentrations were determined simultaneously. A feature extraction algorithm for the algae hyperspeetral data was developed, and was used to estimate Chl-a concentration in aquatic environments. According to the measured data, separate hyperspectral quantitative models for Chlorella vulgaris and Microcystis aeruginosa were developed. Based on the models, Simulate Anneal Arithmetic was applied for the non-linear fitting of the model parameters and achieved the extracted features of the hyperspectral data. The feature extraction algorithm used in this study can effectively predict the hyperspectral characteristics of these two algaes. The relevant Chl-a inversion models for Chlorella vulgaris and Microcystis aeruginos were obtained.