Apple Brown Spot of Hhyperspectral Remote Sensing Estimation Study

In this paper, using hyperspectral remote sensing technology for monitoring Apple made a preliminary analysis of brown spot, through field trials, for diseases of apple leaf spectral data and disease index data. Through comprehensive analysis, several conclusions can be drawn from the following:(1) With the aggravation of the disease brown patch, leaf spectral reflectance in the visible range increases, decreases in the near infrared; leaves first derivative spectrum in the green region of the first-order differential value increases, the red light area decreases.(2) of apple leaf spot disease index and the first derivative data were analyzed. The results showed that the 432 ~ 582nm, 637 ~ 702nm and 715 ~ 765nm wavelength there is significant correlation.(3) Select the correlation coefficient of 0.01 test level of the differential variable, using linear or nonlinear regression estimation technology to build single-variable model of the disease index. For variable D725/D702, Dr, SDr, SDr / SDg, (SDg-SDb) / (SDg + SDb), the best model is the linear model; for variable SDb, the best model for the logarithmic model; For the variable SDr / SDb, SDg / SDb, the best model is the exponential model; for the variables (SDr-SDg) / (SDr + SDg), the best model for the parabolic model; for the variables (SDr-SDb) / (SDr + SDb), the best model is a cubic function model.(4) After testing proved that (SDr-SDb) / (SDr + SDb) for the variables of a cubic function model RMSE estimate the condition index was 5.63%, higher than the accuracy of other models, and the model is less than estimated 10% of the disease index of high accuracy, that it is a leaf disease index of the differential spectral index estimates the best model. Studies suggest that, when the disease index was 5%, the best control of brown spot is Apple, prevention and treatment of apple production is not a big impact. The results of this study showed that the best model is the estimated error was 5.63%, that is, can the use of hyperspectral remote sensing to monitor crop conditions in early disease. Use of hyperspectral remote sensing technology for monitoring the brown spot on the apple, can find diseases at an earlier period, the occurrence and development to provide the status of brown spot, and can generate the spatial distribution of disease incidence and area, while the monitoring can give temporal provide decision makers with important trends in the spread of disease, so that decision-makers targeted decision-making, strengthening the focus on prevention, reducing yield loss. Whether in China or the world apple production, are of great significance.