摘要
【目的】探索快速检测柑橘Vc含量的方法。【方法】利用不同分解水平的Daubechies3小波变换,对100个脐橙整果样品的近红外光谱信号进行了消噪处理,并利用消噪后的重构光谱对脐橙Vc含量进行了偏最小二乘法交叉验证。【结果】小波分解尺度水平不同,偏最小二乘法交叉验证效果各不相同,在分解水平为4时效果最好,其预测值与标准值的相关系数R达到0.9574,交叉验证预测均方差RMSECV仅为3.9mg/100g。比较了11种光谱预处理方法,其中,小波消噪后的偏最小二乘法交叉验证模型预测效果最好,预测值与真值的相关系性最好。【结论】小波消噪后建立的近红外光谱模型能准确地对脐橙Vc含量进行无损快速的定量分析。
[Objective] The objective of this study is to quest for a method to measure vitamin C content of orange. [Method] Based on wavelet transformation by different decomposing levels, the near-infrared spectroscopy signals of 100 intact orange samples were do-noised and some PLS-CV （partial least squared-cross validation） operations were proposed for the prediction of umbilical orange Vc （Vitamin C） content with the reconstructed spectra after do-noised. [Result] The PLS-CV results were not the same when the wavelet decomposing level was different. PLS-CV result was the best at a wavelet decomposing level of 4. Its R was 0.9574, and its RMSECV was 3.9 mg/100g. The 11 different pretreatment methods of Spectroscopy were compared, and the wavelet transformation do-noised PLS-CV results were the best. The correlation of prediction value and true value were the best. [Concluded]The FT-NIR model treated by wavelet do-noised is feasible in prediction of Vc content of umbilical orange rapidly and nondestructively.
出处
《中国农业科学》
CAS
CSCD
北大核心
2007年第8期1760-1766,共7页
Scientia Agricultura Sinica
基金
湖北省科技攻关项目(2004AA101D07)
关键词
脐橙
近红外光谱
小波消噪
偏最小二乘法
Umbilical orange
Near infrared spectroscopy
Wavelet do-noised
Partial least squared
