光谱预处理对苹果可溶性固形物含量VIS/NIR预测模型的影响  

Effect of Spectral Data Pretreatment on Visible/Near Infrared Spectroscopy Model of Soluble Solids Content of Apples

作　　者：刘亚[1] 郭俊先[1] 木合塔尔.米吉提 岳建魁 王艳[1] 刘明涛[1] 

机构地区：[1]新疆农业大学机械交通学院,新疆乌鲁木齐830052 [2]阿克苏职业技术学院,新疆阿克苏843000

出　　处：《北方园艺》2016年第20期1-4,共4页Northern Horticulture

基　　金：国家自然科学基金资助项目(61367001)

摘　　要：可溶性固形物含量是苹果内部品质的重要指标,以"红富士"苹果为试材,测定了从果实膨大定型期到采收期整个成熟阶段,不同生长天数苹果的可见近红外反射光谱;对光谱进行了预处理,包括导数处理、标准正态变换、多元散射校正等方法;利用偏最小二乘法(PLS)建立了苹果可溶性固形物含量的预测模型。结果表明:使用一阶导数和二阶导数处理后的光谱进行预测的准确度高于原始光谱,相关系数分别为0.915 9、0.934 4。The soluble solids content is an important index for apple internal quality.‘Fuji＇apple was used as test material,visible near infrared reflectance spectra of different growth days of apple was measured,during the apple′s ripening stage from enlargement period to harvest period.The original spectrum had pretreated by derivation,standard normal variation transformation and multiplicative scatter correction.Partial least square method was applied to construct the model between soluble solids content and spectrums.The results indicated that the forecast accuracy of the first order derivation and the second order derivation spectrums was higher than that of the original spectrums,the results of the correlation coefficient were 0.915 9,0.934 4.

关 键 词：光谱预处理 苹果成熟期 可溶性固形物含量 可见近红外光谱 偏最小二乘法 

分 类 号：S129[农业科学—农业基础科学]