基于RGB颜色空间的早稻氮素营养监测研究

针对双季稻区水稻过量施肥带来环境污染和成本提高问题,设计不同品种氮肥梯度大田试验,应用数码相机获取早稻冠层数字图像,研究不同色彩参数及早稻氮素营养指标的时空变化特征,以期确立双季早稻氮素营养预测模型。结果表明:不同品种同一氮肥处理下图像色彩参数差异不大;拔节期数字图像参数对氮素营养指标敏感;模型构建结果显示,图像参数INT与水稻氮素营养指标构建的模型决定系数(R2)最大,模型预测效果最佳,R2分别为0.895 7和0.924 7;进一步采用多元回归分析和BP神经网络分析法进行预测,预测效果均较好。对预测结果进行检验,发现品种对于模型的构建影响不大,以BP神经网络分析法构建的叶片氮浓度(LNC)模型和以INT为敏感色彩参数构建的叶片氮积累量(LNA)回归模型效果最优,而多元回归分析方法则效果不佳。早稻冠层RGB颜色空间敏感参数与氮素营养指标间相关性较好,可以实现氮素营养的无损监测诊断。

Monitoring the nitrogen nutrition of early rice based on RGB color space

In view of the problem of environmental pollution and cost increase caused by excessive nitrogen fertilization application of double-cropping rice, field experiments on nitrogen gradients of different varieties were conducted in Jiangxi in 2019. Digital camera was used to measure images of early rice canopy, and different color parameters and characteristics of the nitrogen nutrition indexes of early rice were investigated. A monitoring model of nitrogen nutrition for early rice was established. The results showed that the differences of color parameters of images under the same nitrogen fertilizer treatment of different varieties were small. The digital image parameters at jointing stage were sensitive to nitrogen nutrition indexes. The model construction results showed that the model constructed between INT and the nitrogen nutrition indexes(R2)had the largest coefficient of determination, and the model displayed the best prediction effect with R2 of 0. 895 7 and 0. 924 7, respectively. At the same time, the results of multiple regression analysis and BP neural network analysis confirmed that the prediction results were good. The results obtained from prediction tests showed that the variety has little effect on the construction of the model. The leaf nitrogen content(LNC)model constructed with BP neural network analysis and the leaf nitrogen accumulation(LNA)regression model constructed with INT as the sensitive color parameter had the best effect, while the multiple regression analysis method had poor result. In conclusion, the early rice canopy RGB color space sensitive parameters have a good correlation with nitrogen nutrition indicators, which meets the needs of nitrogen nutrition non-destructive monitoring and diagnosis.