Spectral Identification And Physiological Index Estimation Of Plants Under Stress

In order to explore an effective method to identify whether the plants are under stress or not,this paper takes the commercial crops Cerasus Humilis as an example to conduct the experiment,which is always used for the ecological restoration in the Da Liu Ta Mining in the ShenDong Area.This paper mainly monitors the changes of several parameters of the Cerasus Humilis under the different scales of stress and different kinds of stress by taking the different gradient stress,such as Soil and Plant Analyzer Development(SPAD),Leaf Relative Water Content(RWC)and Chlorophyll Fluorescence.Then,it constructs the spectral characteristic index as the stress diagnostic index which can identify whether the plants are under stress or not effectively basing on the parameters.Finally,it establishes the estimation model of physiological indexes of plants according to the three methods and the imitative effects of the model have been analysed well.Main contents and results are given as follows:(1)After the pre-processing and feature transformation,the spectral curve can effectively highlight the spectral characteristics,the first-order differential transformation and envelope removal of the spectral curve are the best methods,the logarithmic transformation of the inverse spectral value is the second.The experiment devotes to study the relationship between the Soil and Plant Analyzer Development(SPAD),Leaf Relative Water Content(RWC)and Chlorophyll Fluorescence and the spectral curve which was obtained by the method of first-order differential transformation,envelope removal of the spectral curve,the logarithmic transformation of the inverse spectral value respectively,and as a result,it concludes that there are obvious relationship between the physiological index of the plants and reflectance spectrum,which can serve for the analysis of the bands sensitive to the physiological indexes and the relative between each band.The two analytical results can provide the basis for conducting the spectral index.(2)By the continues monitoring of the physiological index of the Cerasus Humilis under stress and spectral curve,the SPAD and RWC are unsuitable for monitoring the plants under stress,whereas,chlorophyll fluorescence parameters can monitor the state of stress effectively,which is the nicer index.In order to monitor the stress state by hyperspectal remote sensing,this paper defined the spectral feature index(NDVI)based on the sensitive bands.According to the distance seperability principle,this index can well distinguish the plants under stress,it can be recognized as the spectral index for the identification of the state of stress,but there are still some drawbacks to estimate the degree and cause of the stress.(3)This paper construct several spectral feature index based on the sensitive bands according to the SPAD and the RWC,it uses the regressive technology such as A simple linear/nonlinear regression,generalized regression neural network and Particle swarm optimization parameters of least squares Support Vector Machine(SVM)to quantitatively model the relationship between the spectral feature index and the SPAD and the RWC.It proves that the three methods are reliable and the method called Particle swarm optimization parameters of least squares Support Vector Machine(SVM)has the highest accuracy,which could be the effective model for inversing the physiological index of plants by hyperspectal remote sensing.