摘要
提出了一种结合主成分分析(PCA)和独立成分分析(ICA)的多光谱数据降维方法,实现了用低维基向量的线性组合来表示高维的光谱数据。首先用PCA方法对多光谱数据进行重构,为了提高色度精度,重建中引入了物体的色度信息;然后用ICA方法对因引入色度信息所造成的剩余光谱误差进行修正。从实验结果来看,所提出的方法均方根误差的平均值较PCA法降低了34.48%,GFC的平均值也达到了95%以上,其降维精度优于基于PCA的光谱降维方法。
A new dimensionality reduction method based on combination of Principal Components Analysis （PCA） and Independent Components Analysis （ICA） is proposed so that high dimensional spectral data can be represented by a linear combination of low dimensional base vectors. Firstly, PCA method is used to spectral reconstruction. Colorimetric information of object is introduced into the spectral reconstruction in order to improve colorimetric accuracy. Then, ICA method is performed to correct residual spectral error resulted from the introduction of colorimetrie information. Experiments results show that the average value of root-mean-square error of proposed method increases by 34. 46 % in comparison with PCA method, and average value of GFC is more than 95%. The accuracy of dimensionality reduction of proposed method is better than PCA method.
出处
《光学技术》
CAS
CSCD
北大核心
2014年第2期180-183,共4页
Optical Technique
关键词
光谱学
光谱反射率重建
主成分分析
独立成分分析
spectroscopy
spectral reflectance recovery
principal components analysis
independent components analysis
