尚静,孟庆龙,黄人帅,易继荣,田驰,张艳.光谱技术对“贵长”猕猴桃表皮颜色的无损检测[J].包装工程,2022,(5):76-82.
SHANG Jing,MENG Qing-long,HUANG Ren-shuai,YI Ji-rong,TIAN Chi,ZHANG Yan.Nondestructive Detection for Color of "Guichang"Kiwifruits by Spectroscopy Technology[J].Packaging Engineering,2022,(5):76-82. |
| 光谱技术对“贵长”猕猴桃表皮颜色的无损检测 |
| Nondestructive Detection for Color of "Guichang"Kiwifruits by Spectroscopy Technology |
| 投稿时间：2021-08-08 |
| DOI：10.19554/j.cnki.1001-3563.2022.05.011 |
| 中文关键词: 光谱技术 “贵长”猕猴桃 多元线性回归 无损检测 |
| 英文关键词: spectroscopy technology “Guichang” kiwifruits multi linear regression nondestructive detection |
| 基金项目:贵州省基础研究计划(科学技术基金)(黔科合基础[2020]1Y270);贵州省科技计划(黔科合基础[2019]1010号);贵州省普通高等学校工程研究中心(黔教合KY字[2016]017);贵阳市科技局贵阳学院专项资金(GYU-KY-〔2021〕);大学生创新创业训练计划(S202010976057) |
|
| 摘要点击次数: |
| 全文下载次数: |
| 中文摘要: |
| 目的 水果表皮颜色是判别其品质的主要指标之一，为实现“贵长”猕猴桃表皮颜色(L*，a*，b*)的快速无损检测。方法 利用光纤光谱仪获得不同成熟阶段“贵长”猕猴桃的光谱反射率;分别运用竞争性自适应重加权算法(CARS)和连续投影算法(SPA)对标准正态变换(SNV)预处理后的全光谱进行降维，进而基于提取的特征变量和测得的表皮颜色值构建猕猴桃表皮颜色多元线性回归(MLR)无损检测模型。结果 对于预测猕猴桃表皮颜色(L*、a*、b*)，采用CARS从1024个全光谱中分别提取了48、48和26个特征变量;相对于构建的SPA-MLR模型，CARS-MLR无损检测模型具有相对较好的校正性能(L*，RC=0.91，RMESC=0.96;a*，RC=0.94，RMESC=0.44;b*，RC=0.92，RMESC=1.08)和预测性能(L*，RP=0.88，RMESP=1.13;a*，RP=0.90，RMESP=0.57;b*，RP=0.90，RMESP=1.15)。结论 基于光谱技术结合多元线性回归预测“贵长”猕猴桃表皮颜色可行。 |
| 英文摘要: |
| The work aims to conduct rapid nondestructive detection for the color (L, a, b) of Guichang kiwifruits, because the color of fruits is one of the major index to estimate the quality. The fiber-optic spectrometer was used to collect spectral reflectance of Guichang kiwifruits at different maturity stages. The competitive adaptive reweighted sampling (CARS) and successive projection algorithm (SPA) were adopted to reduce the dimension of the full spectrum preprocessed by standard normal transform (SNV). Then, the multi linear regression (MLR) model was built up based on selected characteristic variables and the values of the color. 48, 48 and 26 characteristic variables were extracted by CARS from 1024 full wavelengths to predict the color (L, a, b) of kiwifruits. The calibration ability (L, RC=0.91, RMESC=0.96; a, RC=0.94, RMESC=0.44; b, RC=0.92, RMESC=1.08) and prediction ability (L, RP=0.88, RMESP=1.13; a, RP=0.90, RMESP=0.57; b, RP=0.90, RMESP=1.15) of CARS-MLR model were better than SPA-MLR model. Therefore, it is possible to predict the color of “Guichang” kiwifruits based on spectroscopy technology and the multi linear regression. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
|
|
|
渝公网安备 50010702501716号