研究了利用小波分析从浮游植物吸收光谱中提取赤潮藻类特征光谱信息的技术。其中应用了13组单种培养的赤潮藻类的吸收光谱，这些吸收光谱对应的藻类属于4个门，6个属和7个种。首先，对13组吸收光谱数据进行了5级一维小波分析。低频的第5级分量对应于对藻类检测无用且量值较大的背景信息，高频分量的第1—2级为随机噪声，其他高频分量(第3级到第5级)则是对藻类检测有用的信息(或称为特征光谱)。另外，为了检验小波分析在提取特征光谱中的性能，文章利用聚类统计分析方法对提取的13组特征信息进行了聚类。聚类结果显示，13组吸收光谱被正确地进行了分类，这验证了小波分析在提取藻类特征光谱中的有效性。

In the present study, the feasibility of using wavelet analysis to extract the eigen spectra from the absorption spectra of phytoplankton for species detection was investigated. Thirteen absorption spectra taken from single species cultures, representing four divisions (Dinophyta, Bacillariophyta, Haptophyta, and Chlorophyta), six genus (Gymnodinium, Prorocentrurn, Skeletonerna, Guinardia, Phaeoc ystis, and Prasinoph yte ) and seven species ( Karenia mikirnotoi, Karenia brevis, Prorocentrum minimum, Skeletonema costatuma, Guinardia delicatula, Phaeocystis globosa, and Pyramimonas parkeae ), were used. First, the 1D wavelet analysis with five levels was applied to the thirteen absorption spectra, so each spectrum was decomposed with 5 levels. The 5th level component of low frequency corresponds to the background without information for species detection, and 1st and 2nd level component of high frequency is the random noise. The other levels (3rd to 5th) of high frequency are the useful information, and the sum of levels (3rd to 5th) of high frequency was retained as the eigen spectra for species deteetion Second, the clustering analysis was used to the eigen spectra for examining the performance of the wavelet analysis in extracting species information. The clustering results were compared with the known species class information, and the results show that the 13 absorption spectra are correctly classified at the level of division, genus and species. This means that the wavelet analysis has good performance in extracting the eigen spectra for species detection. However, the above results were obtained with only limited species, and the more species data are required to identify the extensive validity of the conclusion.