Studies On Ginkgo Biloba Leaf Chlorophyll Content Estimation Based On Image Analysis And Hyper-spectral Analysis

In order to predict the chlorosis of Ginkgo biloba leaves at the early stage and establish a diagnostic system,support the long-term dynamic observation of chlorophyll content,provide a new method for the management of garden green space,the image method and hyper-spectral analysis method,two kinds of plant physiological information nondestructive testing methods with the rapid development in recent years,were used to measure the chlorophyll content of Ginkgo biloba leaves.Through the optimization of the image programming code,the construction of the spectral sensitive bands and the vegetation index,the optimal model was established and he prediction accuracy was improved.The following findings were obtained from two technical perspectives:(1)Image methodThrough the code optimization of the image processing process,the image enhancement method-the median filter method(5×5 window)and the image segmentation method-histogram threshold segmentation method,which were suitable for the color extraction of the scanned image,were selected to realize the color image to the rapid extraction of color values.Among the 44 color indicators,the optimal color model of chlorophyll content in Ginkgo biloba leaves was R-B index model,the expression was y=3.142e-0.041x(R2=0.767,estimated accuracy=97.16%).Therefore,R-B index model can be used as the best predictor of chlorophyll content estimation based on image processing.(2)Hyper-spectral analysis methodThe sensitivity of the original spectral spectrum of Ginkgo biloba leaves were 605nm and 700nm,and the sensitivity bands of the first derivative spectra were 551nm,655nm,690nm and 724nm,and the correlation between chlorophyll content and chlorophyll content in the three-sided parameters was analyzed by the selection of spectral sensitive bands and the vegetation index.The parameters with high correlation and chlorophyll content in the three edge parameters were(SDr-SDy)/(SDr + SDy),(SDr-SDb)/(SDr + SDb),Dy,and the higher correlation between chlorophyll content and vegetation index were(R640-R673)/R673?(R675×R690)/(R683×R683)?R605/R760.The optimal spectral characteristic parameter of chlorophyll content of the leaves DR551 was select from Spectral sensitive band,Three-side parameters and egetation indexes with the estimated equation y=1247976.043x2-3149.lx+2.7052(R2 = 0.865).In addition,the comprehensive vegetation index(R675*R690)/(R683*R683)showed better correlation with logarithm,exponential and linear equations(R2>0.850,estimated accuracy>95%).Therefore,both the first derivative 551nm and the comprehensive vegetation index(R675*R690)/(R683*R683)can be used as the estimation index to detect the chlorophyll content of Ginkgo biloba leaves.