Research On Detection Method And Application Of Plant Chlorophyll Fluorescence Kinetic Imaging

One of the most important task and challenge of our country is to promote the sustainable development of agriculture,develop the fine agriculture and ecological agriculture.It is necessary to strengthen the application of automatic measurement technology in agriculture,especially for plants,the demand of water and nutrients and disease early detection.Chlorophyll fluorescence is closely related to the photosynthesis of plant PS ?,participate in the competition for energy distribution,is known as the nondestructive probe of plant testing.The number and structure of Chlorophyll molecule will change greatly when plant under the states of water stress,nutrition stress,disease stress,this change could be detected sensitively by Chlorophyll fluorescence before symptoms appear.Compared to transmitted light,the use of Chlorophyll fluorescence to detect can effectively avoid the measurement error caused by different leaf thickness.Therefore,in recent years,Chlorophyll fluorescence is widely used to detect plant biotic and abiotic stresses.In this thesis,water stress,water and fertilizer coupling and disease stress were studied by imaging analysis of Chlorophyll fluorescence.According to the change characteristic of Chlorophyll fluorescence kinetics curve,the Chlorophyll fluorescence imaging system was proposed and designed,consisting of LEDs for an excitation at 460 nm,an EMCCD(Electron-Multiplying CCD)detector,filters,lens,lifting platform etc.Kinetic parameters on fast rising part and slow falling part of the Chlorophyll fluorescence kinetic curve,as well as the Chlorophyll fluorescence images at different moments could be detected and obtained.This system can provide experimental support for the two-dimensional observation of fluorescence images of plant and the qualitative and quantitative analysis of plant growth state by using chlorophyll fluorescence kinetics parameters.To solve the problem that the measurement error of Chlorophyll content in plants is greatly affected by water content and thickness of leaves when Chlorophyll content is represented by SPAD(Soil and Plant Analyzer Development)value,an online in-situ Chlorophyll content and moisture content measurement system of living plant was proposed and designed,and a fine Chlorophyll content detection model based on BP neural network was constructed.The mode was verified by taking three species plants as samples,such as bauhinia,Photinia serrulata and holly tree leaves.The results showed that compared to the traditional single-parameter(SPAD)model,the measurement accuracy improved greatly under the BP neural network model with consideration of factors,such as leaf water content and thickness.Especially for plants with thick leaves,this model can effectively improve the accuracy of online detection.To qualitative analysis about the change of Chlorophyll fluorescence images and Chlorophyll fluorescence kinetics parameters when different plants under water stress,sample leaves were measured at four different rates using the designed system in this thesis.Fluorescence images were operated with noise reduction,multiplication,division and pseudo color display,the change of plant Chlorophyll fluorescence distribution caused by various factors could be intuitive analyzed and the chlorophyll fluorescence kinetics curve with more information was obtained and the kinetics parameters were calculated.A method for determining the water stress of plants was proposed by using fluorescence ratio Rfa,TFM(appear time of secondary peak in kinetics curve)and other effective parameters.This qualitative research provides research foundation and characteristic parameters for further quantitative analysis of plant growth and physiological state and mathematical modeling.Water and fertilizer are promoted and restricted with each other during the process of plant growth,in this thesis,plant water-fertilizer coupling model were established using Chlorophyll fluorescence kinetics parameters.Samples with different water treatments,different amount of nitrogen fertilizer and different water-fertilizer coupling level were detected and experimented.The features were extracted and the model were established,recognition accuracy of different modeling methods had been compared.The results showed that plant under different water-fertilizer coupling levels could be discriminated by the model established base on RBF(Radial Basis Function)neural network.With respect to plant disease,the early recognition and early warning of cucumber wilt disease was studied based on Chlorophyll fluorescence in this thesis.The gray GM(1,1)mode was established using parameter Rfd,healthy and disease leaves could recognized about eight days after inoculation using this model.The model of cucumber wilt disease was established based on RBF neural network,which using characteristic values both on fast phase portion and slow phase portion of Chlorophyll fluorescence kinetics curve,it effectively improves the advance recognition accuracy.The experimental results showed that the accuracy of identifying cucumber wilt disease on the fourth day increased from 73.3%to 82.2%.Through the theoretical and experimental research,the results showed that,the Chlorophyll fluorescence kinetics image detection technology could be used to effectively detect the water stress,nutrition stress,disease stress and other environment stresses.The RBF model with both fast phase and slow phase feature points could effectively increase the recognition accuracy of plant stress type and stress level.The study in this thesis provide a new auxiliary detection method on the development of precision agriculture and the research of the complex mechanism on multi-factors react each other during the growth process of plants.