摘要
利用Savitzky-Golay(SG)卷积平滑方法和标准正态变量变换(SNV)对土壤光谱数据进行处理后,通过对波段的优选建立了主成分回归(PCR)模型。结果表明,预测样本的相关系数可达到0.932 2,预测标准差(RMSEP)为0.041 1%。
The pretreatment methods of Savitzky-Golay and Standard Normal Variate Transformation（ SNV） are used to deal with the soil spectrum. In order to set up a better prediction model,waveband optimization is necessary. This paper presents a new way to make the waveband optimization. Meantime,the prediction model is built using the Principal Component Regression（ PCR）. The result shows that the correlation coefficient can reach 0. 932 2 and the standard error of prediction is 0. 041 1%.
出处
《仪表技术》
2014年第5期4-6,14,共4页
Instrumentation Technology
关键词
土壤有机质
近红外光谱
波段优选
主成分回归
soil organic matter
near infrared spectroscopy
waveband optimization
PCR
