基于小波变换的藻类显微光谱成像分类技术

通过显微成像光谱技术对淡水藻类进行成像光谱的测量,并试图同时从形貌和化学成分两个方面给出藻类识别的综合方法。首先利用显微图像技术对单个藻种进行观察测量,进行形貌学分类,再通过成像光谱仪获得藻类的吸收光谱立方体,然后利用小波变换对光谱进行处理,得到区分形貌相近的不同藻类的光谱特征参数,在化学成分上将其区分。本方法对准确进行藻类识别具有重要的指导意义。

Technology of Algae Identify by Absorption Spectral Imaging Microscopy Based on Wavelet Transform

The concurrency frequencies and distribution areas of the water blooms in fresh water are increasing in China.It is not only destroying natural aquatic ecosystems,but also damaging human health.So,the study of this field is popular in recent years.To control effectively the water blooms,the important element is identifying the algae's species in the water blooms.By the coherent papers,we know the main species are blue algae,green algae and diatom etc.We develop a more effective way to identify the algae in the water blooms.Combined the microscopic imaging technology and spectrum imaging technology,we can use microscopic imaging technology for pattern classification firstly,and then use spectrum imaging technology for chemical separation of algae.Through the establishment of algae spectral cube by imaging spectrometer,which can be arbitrary pixel of the image spectrum.Two typical of the species selected for absorption spectroscopy-microcystis aeruginosa(blue algae) and chlorella pyrenoidosa(green algae).Different algae species have different pigments and different pigments have different absorption peak,so that distinguish easily between different species by absorption peak.But in the experiment,we abtained the absorption spectra superimposed of all the material in the algae.It is quite difficult to identify directly through the absorption spectra.So,we processed the spectra data using wavelet transform tools of Matlab 7.0.Wavelet transform can decompose signals into localized contributions of different frequency,it is a effective time-frequency analysis tool.Some decomposed components with some high frequency have the possibility to separate different absorption peak.By related papers and some experiences,we chose appropriate component to analysis the data.Experimental results show that the method was feasible.So,this paper provides a scientific and effective recognition way.It is of important guiding significance to prevent the water blooms.