基于高光谱成像技术的草莓硬度预测  

Prediction on Firmness of Strawberry Based on Hyperspectral Imaging

作　　者：卢娜[1,2] 韩平 王纪华 

机构地区：[1]三峡大学计算机与信息学院,湖北宜昌443002 [2]北京农业质量标准与检测技术研究中心,北京100097

出　　处：《软件导刊》2018年第3期180-182,共3页Software Guide

摘　　要：为对草莓硬度进行预测研究,利用高光谱成像系统获取草莓的高光谱数据,光谱数据波长为400~1 000nm,采用标准正态变换(SNV)、多元散射校正(MSC)、卷积平滑方法(Savitzky-Golay)以及几种方法相结合对光谱数据进行预处理,选择最优的预处理方法,进一步结合化学计量学方法建立PLS预测模型,比较不同的光谱预处理方法对预测模型的效果,以选择最优预测模型。结果表明,经标准正态变换(SVN)处理后建立的偏最小二乘(PLS)模型效果最好,校正集和预测集的相关系数及均方根误差分别为0.989,0.882和0.021,0.073。因此,可采用高光谱成像技术对草莓硬度进行预测。To predict the firmness of strawberry, hyperspectral data of strawberry were obtained by hyperspectral imaging system. The spectral data were wavelengths of 400- 1 000nm. There were used to acquire the best pretreatment method in the spectral region that is standard normal transform （SNV）, multiple scattering correction （MSC）, convolution smoothing method （Savitzky-Golay） and combined several methods. We will establish the partial least squares forecasting model with chemometrics and then compare the effect of different spectral preprocessing methods on the prediction model to select the optimal prediction model. The results show that the partial least squares （PLS） model established by standard normal transform （SVN） is best. The correlation coefficients and root mean square errors of the calibration set and the prediction set are 0. 989, 0. 882 and 0. 021, 0. 073, respectively. Therefore, hyperspectral imaging techniques can be used to predict the firmness of strawberry.

关 键 词：高光谱成像技术 草莓 硬度 无损检测 

分 类 号：TP319[自动化与计算机技术—计算机软件与理论]