摘要
快速、精确、无损地检测脐橙叶片氮(N)素含量,对脐橙果树N肥施用、果树增产有重大现实意义。用正自适应加权算法(CARS)、遗传算法(GA)、连续投影算法(SPA)提取脐橙叶片高光谱图像的有效信息,对全氮含量用偏最小二乘法(PLS)、多元线性回归法(MLR)、主成份回归法(PCA)进行建模定量分析。高光谱图像标定后,提取感兴趣区域(ROI)的平均光谱,选出敏感波段进行建模。结果表明:GA—MLR模型具有较高的优势,预测模型相关系数R=0.82,均方差RMSEP=0.32。应用高光谱技术结合此模型可以对对赣南脐橙叶片全氮含量快速、无损的定量分析。
It has great practical significance to fast,accurately and non-destructively detects nitrogen(N)pigment content in Gannan Navel Orange for N fertilizer and fruit production.Competitive adaptive reweighted sampling(CARS),Genetic Algorithms(GA),Successive Projections Algorithm(SPA)and other methods were used to extract navel orange leaves hyperspectral images valid information.Also,Partial Least Squares(PLS),Multiple Linear Regression(MLR)and Principal Component Regression(PCR)were used for modeling quantitative analysis.Average spectral was extracted from region of interest(ROI)of hyper-spectral images after pre-processing.Finally,sensitive bands were selected for modeling.The results showed that:GA—MLR model has more advantage with correlation coefficient Ris 0.82,MSE RMSEPis 0.32 in prediction model.Hyperspectral technology combined this model can analysis leaf nitrogen content in Gannan navel orange leafs fast and non-destructively.
出处
《中国农机化学报》
2016年第9期99-103,共5页
Journal of Chinese Agricultural Mechanization
基金
国家863计划课题(2012AA101906)