| Microalgae not only contain protein,amino acid,un-saturated fatty acid and many kinds of pigment,but also it is valuable as raw material to produce ethanol and alima,natural pigment,and ‘cleaner’ for wastewater in commercial area.Today the most challenging matter is improving microalgae growth rate in product synthesis.In order to maximize the commercial value of microalgae as raw materials for natural pigment extraction and health products,as well as bioenergy,further observation and research on the internal active substances such as oil and other substances in algal cells were carried out to realize rapid system for automation analysis.lay the foundation for the formulation of efficient production strategies.Hyperspectral imaging technology have the spectral resolution of several nanometers,which combined with optics,electronics,information processing and computer science technology.At the same time,it organically combines traditional two-dimensional imaging technology and spectral technology to make it have the advantages of high spectral resolution and conjoining spatial and spectral information.Spectrum is a kind of pattern which can reflect the chemical and physical properties of the substance to be tested,and cell image is an intuitive embodiment of cell morphology information.In this paper,firstly,the hyperspectral image of microalgae is obtained by the fluorescence hyperspectral microscopic system.secondly,the classification of dominant algae species is constructed and finally the analysis of pigment components in the cells of the algae species is carried out.Fluorescence hyperspectral microscopic system is constructed by light source,optical system,hardware module and detector.Hardware system mainly improves the imaging resolution by electrically controlled high precision displacement table,and ensures the spatial resolution around 1.29μm under the premise of high spectral resolution.The key technologies of system is the step-control problem of motorized stage and the synchronization of scanning device and spectrometer,Through self-writing program to achieve accurate control and solve synchronization problems.Taking Microcytic aeruginosa,Anabaena,Chlorella,Chlamydomonas reinhardtii,,haematococcus pluvialis as sample,the hyperspectral data is obtained.In the prepossessing,using the principal component analysis and successive projection algorithm to dislodge redundancy in the data.The popular classifier such as support vector machine and extreme learning machine is used to classify microalgal cells that the data feature is obtained by prepossessing.Compared to absorbance spectrum,it induced classification accuracy of hyperspectral data,reaching by 96.67%.This result is not only recognize the blue-algae and the green algae,but also it improves the classification of green-algae.The specific microalgae under extreme condition may accumulate lipid and produce several new pigments.According to this,the hyperspectral data of Chlorella is obtained HSI for a whole cell circle.A PLS model was established based on these data to predict the growth status of Chlorella by the relationship between the proportion of oil content and the days.In general,by using HSI,the information of pigment and oil of microalgae life information was obtained,and the distribution and content of pigment in microalgae cells were obtained quickly.Besides,a model for prediction of growth trends was obtained.It is of great significance to provide a reliable and rapid monitoring method for the growth of microalgae and to formulate feasible production strategies for later industrial production. |
PDF Full Text Request