Prediction Of Maize Phenotypic Characteristics Based On RGB Image Method

Phenomics is a subject that mainly excavates the dominant phenotypes of organisms under different external environment and physiological characteristics.However,due to the variability and instability of the environment itself and the complex monitoring of the biological internal physiological characteristics,it is impossible to grasp the dynamic changes in real time,so the traditional research basically stays in some obvious phenotypic indicators,such as plant height and stem diameter.However,this kind of research can not completely generalize the specific characteristics of phenotype,which is quite general.The research itself is only a rough correlation analysis of these indicators,lacking the mining of new indicators and the exploration of core indicators that can show the overall characteristics of the organism,and temporarily also can not fully express the single phenotypic characteristics of the dominant characteristics of the organism At the same time,because of the low efficiency of the traditional monitoring methods,and the subjective factors among the investigators,and because of the long period of phenotypic monitoring on crops and the slow popularization of phenotypic technology,the phenotypic monitoring methods are far behind the current situation of theoretical research and even slow down the development rate of phenotypic research technology.Therefore,in this experiment,the phenotypic group method and technology,combined with the data of the phenotypic characteristic indexes of Maize in visible light,were used to extract some core indexes to reflect the characteristics of multiple morphological characteristic values and colors of multiple varieties in different environments and different periods of time.The phenotypic monitoring index model and sampling method based on visible light imaging were established Based on the science of quantification and indexation,it can provide low-cost and high-efficiency phenotypic detection technology for large-scale crop variety screening,and at the same time,it can promote the application of Phenomics in the field of modern agriculture.In this paper,through the study of RGB color eigenvalues,it is found that the R values of RGB,R,G and B are all greater than 0.9,which are strongly correlated with each other,while the correlation between these basic values and other features is at a low level.At the same time,except RGB,R,G,B,other features have no strong correlation with SPAD,so the constructed SPAD and color feature value are single factor models.The correlation coefficient between pixel proportion and leaf area index is 0.92,which is strong correlation.The correlation coefficient between RGB and SPAD is0.75,so the pixel index model is constructed.Finally,the combination of yield prediction model and "average pixel ratio +RGB + plant height + stem diameter" is the highest by using the possible function in R language.This combination appears in the fifth stage of the big bell mouth period,and the correlation coefficient of each model combination in the fifth stage of the big bell mouth period is higher than that in other periods,which is the best period to observe and predict yield.If the yield is y,the pixel proportion is a,the RGB value is B,the plant height is C,and the stem diameter is D,then the estimated yield formula is y =-186.5 + 232.6a-0.5b + 1.6c + 13.4d.Through the experiment of visible image method,this paper summarizes the method of RGB image analysis and prediction of common phenotypic data,and produces the linear regression equation of estimating common phenotypic data such as the height and leaf area index of corn stem and coarse plant by RGB parameter and pixel parameter,and through the auxiliary combination of corn common index and RGB image parameter,it is related in the fifth stage of corn experiment It has the highest quality,so it can predict the yield of corn in this period;and the yield formula is of great significance to develop the rapid monitoring of corn phenotype and yield prediction,and it can provide reference for the monitoring of the growth of corn crops in the whole province and even in the whole country.