摘要
355nm激光器发射激光入射到海水表面,激发海表面溢油的荧光光谱,运用高光谱图像降维中应用广泛的分段主成分分析算法对油荧光光谱进行波段选择。该算法把每个分段被映射到主成分的信息量的大小作为是否被选择的标准,保证了选择波段的信息丰富;通过分段分析消除了传统主成分分析的全局性引起的波段忽略问题,获得较为满意的降维效果。
The laser with the wavelength of 355nm irradiates to the sea surface, and excites the fluorescence spectrums of oils spilled on the sea surface. The segmented principal component analysis （PCA） algorithm widely used in the dimensionality reduction of hyper spectral image, was used to band selection according to oil fluorescence spectrums. This algorithm regards the amount of the information that ;s mappen into the given band as the selected criterion, it can ensure that the selected bands contain most principal components of a information of the original data. With the segmentation on original data, it can eliminate the problem of band neglect which was caused by global transform in traditional PCA algorithm. And the result of dimensionality reduction is advisable.
出处
《地理空间信息》
2009年第3期12-15,共4页
Geospatial Information
基金
国家863计划资助项目(2006AA06Z415)
关键词
激光油荧光光谱
分段主成分分析
波段选择
laser oil fluorescence spectrums
segmented principal component analysis
band selection
