基于支持向量机回归预测水稻叶片SPAD值

Prediction of SPAD Values of Rice Leaf Based on Support Vector Regression

提出了一种基于机器学习的水稻叶片SPAD值预测方法。通过水稻栽培实验获取样本叶片的SPAD值和叶片图像的RGB值,应用支持向量机原理,建立以叶色图像RGB值为输入参数,叶片SPAD值为输出参数的回归模型。通过样本训练测试和预测实验,对水稻叶片SPAD值预测结果的平方相关系数为91.70%,平均相对误差为3.423%。结果表明支持向量机回归模型对水稻叶片SPAD值的有很好的预测结果,能够满足农学研究的要求,研究方法具有良好的普适性和推广性。

A SPAD value prediction method for rice leaf based on machine learning was proposed.The SPAD value of leaf and the RGB value of leaf image were obtained by rice cultivation experiment, the model based on SVR was established with the RGB value of the leaf color image as the input parameters and the SPAD value as the output parameter. Through the training and for ecastinge xperiments, the squared correlation of SPAD prediction is 91. 70% and the average relative error is 3. 423%. The results showed that the support vector regression model had a good prediction of the SPAD value and could meet the requirement of agronomic research. The method has good universality and generalization.