论文部分内容阅读
In order to establish a fast and non-destructive estimation model of chlorophyll content in maize leaves, 2-CCD multi-spectral images of maize canopy were analyzed by using image processing technology.In this experiment, the images of corn canopy in jointing stage were captured by the 2-CCD multi spectral image intelligent sensing system which was developed independently in the lab.To eliminate the influence of different illumination conditions and background soil on the quality of image acquisition, the image data of the leaf area under different illumination conditions was converted to the leaf reflectance data.In order to improve the accuracy of the algorithm, a piece of four different levels diffuse reflection standard gray plate which meet the conditions of Lambertian was designed and real time calibration.And Image of diffuse reflection Standard Board was collected synchronously, and the NIR, G, B and R of the standard plate image are obtained respectively, and its average gray values of four bands(R, G, B, and NIR) werecalculated as a reference value for color correction.The linear empirical model of the reflectance of the diffuse reflection standard plate and the average gray values was obtained.The canopy image was smoothed and filtered in jointing stage of Maize and divided based on HSI color space.The average gray values of four bands(R, G, B, and NIR) were extracted from the canopy image.The reflectance of four bands(R, G, NIR, and B) was obtained by the linear empirical formula.Leaf reflectance data were measured by spectrophotometer and the average reflectivity was calculated by the spectral reflectance extracted from 2-CCD center wavelength, the correlation between the average reflectivity and the image acquisition was analyzed.4 kinds of common vegetation indices (DVI, NDGI, RVI and NDVI) were calculated respectively by reflectance which obtained by linear empirical model and the average gray values calculated by image processing as the image parameters.The correlation between the parameters of image detection and the content of chlorophyll was analyzed.A multiple linear regression (MLR) model for chlorophyll content was established.Research results show that: the effect of different illumination conditions and background on the image acquisition quality was eliminated because of the diffuse reflection calibration board.There was a high correlation between the reflectivity of the blade and the image acquired by the spectrophotometer.Thus compared with the correlation between vegetation indices calculated by the average gray value and chlorophyll content, the correlation between vegetation indices calculated by reflectance and chlorophyll content has been improved.It is proved that the method is feasible to establish the model of the reflectance of maize canopy leaf image by the diffuse reflection standard board, and the results provided support for the nondestructive testing of the chlorophyll content in Maize Jointing Stage.