Estimation Of Chlorophyll And Phosphorus Contents For Apple Leaves Based On Hyperspectral And Image Technology

Chlorophyll and phosphorus are the key elements for the tree growth.Traditional methods of obtaining P-content and Chlorophyll content by chemical methods require leaves to be grinded and chemical titrations,which would be harmful to apple trees and time-consuming,though the results were accurate.Hyperspectral remote sensing can detect the successive spectral curve of ground objects with hyperspectral resolution in some spectral region and form the unique multi-dimensional spectrum space,which is widely used in all kinds of plant diseases and insect pests′prevention,chlorophyll content prediction,crop yield forecast and crop nutrient elements monitoring etc.Digital image processing technology emerges in response to the needs of times,which provides a scientific basis for diagnosis of the apple tree physiology.In short,the application of hyperspectral remote sensing and image technology can estimate the tree leaves phosphorus contents and chlorophyll contents and to increase production efficiency and the utility ratio of nutrients,which would improve the scientific management of apple trees orchards.strengthen the monitoring ability of physiological ecological parameters and improve the monitoring accuracy of crop growth.The data were from new shoots prosperous long-term Red Fuji apple orchards in Qixia of Yantai City and Mengyin of Linyi City,which included P-content and chlorophyll content of the apple tree leaves.The original spectral reflectivity data were captured by the ASD Field Spec 4 hyperspectral spectrometer and the image were taken through the digital camera.The P-content and chlorophyll content were measured by the traditional chemical analysis in laboratory.In order to increase the accuracy of P-content estimation for apple trees,the data was transformed through the first derivative of original spectral reflectivity,vegetation index and hyperspectral characteristic parameters.According to the result of systematic analysis and diagnosis,Random forests models were established for the P-content.In order to increase the accuracy of chlorophyll content estimation for apple trees,the color parameters were combined based on the RGB values which were extracted from histogram in Photoshop.CS6.0(Adobe System,Inc.).Then the correlation analysis method was used to select the sensitive color parameters.Compared to the trichromatic color values,the combination of RGB values considerably improved the correlation coefficients.The effect of the SVM model was established for the chlorophyll content of apple trees.The main results are as follows:(1)The most sensitive spectral bands were screened of phosphorus content in apple leaves.By the correlation analysis,the P-content negatively correlated with original spectral reflectivity(350~2500 nm range)as a whole,and the correlation coefficient reached to remarkable level,especially in the green light region(507~590 nm),red light region(694~743 nm)and near-infrared spectrum(1324~1364 nm).The highest correlation coefficient was r =-0.6485 at R1720.(2)The core color parameters with different chlorophyll contents from image of apple leaves were screened.Compared to the trichromatic color values,the combination of RGB values considerably improved the correlation coefficients.Among the color parameters,B、 B/R、G、 B/G、G/(R+G+B)、B/(R+G+B)、(RB)/(R+B)、(G-B)/(G+B)、(R-B)/(R+G+B)and(G-B)/(R+G+B)values were correlated significantly with chlorophyll content respectively.(3)The estimation models of phosphorus content for apple leaves were established.By the analysis,the random forest model based on remarkable wavelengths from the vegetation index(DVI(556,712),DVI(677,1728),RVI(542,1094),RVI(705,937),NDVI(937,549),DVI(FDR567,FDR1980)and DVI(FDR523,FDR1883))had the best estimation effect,including the determination coefficient R2=0.9236,root mean square error RMSE=0.0158 and relative error RE=6.9150%.(4)The estimation models of contents for apple leaves were established.The support vector machine(SVM)models were built based on the sensitive color parameters(B、 B/R、G、 B/G、G/(R+G+B)、B/(R+G+B)、(R-B)/(R+B)、(GB)/(G+B)、(R-B)/(R+G+B)and(G-B)/(R+G+B))for different chlorophyll contents(Chl.a、Chl.b、Chl.(a+b)and SPAD,including the determination coefficient R2 were 0.8754、0.8374、0.8671 and 0.8129,root mean square error RMSE were 0.0194、0.0350、0.0497 and 0.9281,relative error RE were 0.8059%、1.7540%、1.122% and 1.1894%.Especially the model for Chl.a had the best estimation.