Monitoring green tea fixation quality byintelligent sensors: comparison of imageand spectral information
智能传感器监测绿茶发酵质量：图像和光谱信息的比较
- 关键词：
- 来源：
- Journal of the Science of Food and Agriculture 期刊
- 类型：
- 学术文献
- 语种：
- 英语
- 原文发布日期：
- 2022-11-23
- 摘要：
- BACKGROUND：Intelligentmonitoring of fixation quality is a prerequisite for automated green teaprocessing. To meet the requirements of intelligent monitoring of fixationquality in large-scale production, fast and non-destructive detection means areurgently needed. Here, smartphone-coupled micro near-infrared spectroscopy anda self-built computer vision system were used to perform rapid detection of thefixation quality in green tea processing lines. RESULTS：Spectral and imageinformation from green tea samples with different fixation degrees werecollected at-line by two intelligent monitoring sensors. Competitive adaptivereweighted sampling and correlation analysis were employed to select featurevariables from spectral and color information as the target data for modeling,respectively. The developed least squares support vector machine (LS-SVM) modelby spectral information and the LS-SVM model by image information achieved thebest discriminations of sample fixation degree, with both prediction setaccuracies of 100%. Compared to the spectral information, the imageinformation-based support vector regression model performed better in moistureprediction, with a correlation coefficient of prediction of 0.9884 and residualpredictive deviation of 6.46. CONCLUSION：The present studyprovided a rapid and low-cost means of monitoring fixation quality, and alsoprovided theoretical support and technical guidance for the automation of thegreen tea fixation process.
- 所属专题：
- 60
