基于冠层数码图像的玉米长势和氮素营养状态分析

Analysis of Maize Growth and Nitrogen Nutrition Status Based on Digital Camera Image

5个施氮水平的田间试验条件下,在玉米9叶期前,分3次采集玉米冠层图像,同步进行破坏性取样,测定玉米地上部干物质质量、叶面积指数和地上部氮积累量;分析了通过玉米冠层图像分析软件计算的冠层覆盖度和10种色彩指数与玉米地上部干物质质量、叶面积指数和地上部氮积累量的关系,以逐步回归和非线性回归方法,构建了玉米长势估算模型。结果表明,玉米地上部干物质质量、叶面积指数和地上部氮积累量与冠层覆盖度和大部分色彩指数间呈显著相关,其中冠层覆盖度与长势指标间的相关性最强,且大部分色彩指数与长势指标间呈非线性相关。通过逐步回归方法和非线性回归方法建立的两个系列模型均能较准确地预测玉米地上部干物质质量、叶面积指数和地上部干物质质量,以逐步回归方法建立的估算模型的预测效果,除地上部干物质质量的预测值与实测值的回归系数外,其余指标均好于基于冠层覆盖度的非线性回归模型,但差异较小。

Digital images of maize canopies grown under five level of nitrogen treatment were taken with a digital camera for three times before 9-leaf stage and at the same time four plants were sampled to measure shoot dry weight, leaf area index, and shoot nitrogen accumulation in 2013. Canopy cover and 10 color indices were calculated from corn digital images by image analysis program （developed in Microsoft Visual Basic）. Correlation analysis were carried out to identify the relationships between canopy cover, 10 color indices and corn growth and nitrogen nutrition parameters, such as leaf area index, and shoot nitrogen accumulation etc. Stepwise multiple linear regression method and nonlinear regression method were used to establish the models to determine maize growth and nitrogen status. The results showed that canopy cover and all color indices except r and Hue showed significant correlations with shoot biomass, leaf area index, and shoot nitrogen accumulation. Canopy cover revealed the highest correlations with maize growth and nitrogen nutrition parameters. Canopy cover and more than 8 color indices were nonlinearly related to maize growth and nitrogen nutrition parameters. Stepwise multiple linear regression models, constituting canopy cover and one color index, and nonlinear regression models with the independent variable of canopy cover were constructed to estimate shoot biomass, leaf area index, and shoot nitrogen accumulation. The validation results indicate that two series regression models had good ability to estimate shoot dry mater, leaf area index, and shoot nitrogen accumulation using canopy cover and color indices of color digital camera image. The validation results also indicate that the models using stepwise multiple regression method had better performance compared with the nonlinear regression model with independent variable of canopy cover.