形状平滑约束在多元曲线分辨交替最小二乘(MCR-ALS)解析高光谱成像中的应用

Multivariate Curve Resolution Alternating Least Squares with Shape Smoothness for the analysis of Hyperspectral Imaging

高光谱成像在多学科研究中提供了丰富的数据信息,由于数据量庞大,研究人员使用化学计量学方法对这些数据的信息进行提取。多元曲线分辨交替最小二乘(MCR-ALS)方法能够分辨混合体系的高光谱数据中的纯组分对应的光谱和浓度信息,得到了广泛的使用。为了进一步提高MCR-ALS对高光谱的解析能力,本文使用了形状平滑约束(SSC)分别分析了模拟数据和实验数据,结果表明,通过形状平滑约束,能够进一步提高MCR-ALS对高光谱数据解析的准确度,而且使MCR减少了扭曲模糊,在二线性分辨中获得了唯一解。

Hyperspectral imaging have been widely applied in the research of biology,agriculture,ecology,environment and food industry.Multivariate Curve Resolution Alternating Least Squares(MCR-ALS)can be used for the resolution of hyperspectral imaging and provide the information of pure spectra and concentration distribution in the mixture systems.MCR-ALS can facilely apply different constraints depending on the feature of the analyzed systems.In this work,MCR-ALS with Shape Smoothness Constraint(SSC)was used to resolve two hyperspectral imaging data from simulation and experiment.The results showed that MCR-ALS with non-negativity constraint and SSC gave unique results without rotation ambiguity,which fitted the raw data better than MCR-ALS with only non-negativity constraint.