• EI
  • Scopus
  • 食品科学与工程领域高质量科技期刊分级目录第一方阵T1
  • DOAJ
  • EBSCO
  • 北大核心期刊
  • 中国核心学术期刊RCCSE
  • JST China
  • FSTA
  • 中国精品科技期刊
  • 中国农业核心期刊
  • CA
  • WJCI
  • 中国科技核心期刊CSTPCD
  • 中国生物医学SinoMed
中国精品科技期刊2020
金山峰,王冬欣,黄俊仕,等. 基于计算机视觉的茶叶品质在线评价系统[J]. 食品工业科技,2021,42(14):219−225. doi: 10.13386/j.issn1002-0306.2020100152.
引用本文: 金山峰,王冬欣,黄俊仕,等. 基于计算机视觉的茶叶品质在线评价系统[J]. 食品工业科技,2021,42(14):219−225. doi: 10.13386/j.issn1002-0306.2020100152.
JIN Shanfeng, WANG Dongxin, HUANG Junshi, et al. Online Evaluation System of Tea Quality Based on Computer Vision [J]. Science and Technology of Food Industry, 2021, 42(14): 219−225. (in Chinese with English abstract). doi: 10.13386/j.issn1002-0306.2020100152.
Citation: JIN Shanfeng, WANG Dongxin, HUANG Junshi, et al. Online Evaluation System of Tea Quality Based on Computer Vision [J]. Science and Technology of Food Industry, 2021, 42(14): 219−225. (in Chinese with English abstract). doi: 10.13386/j.issn1002-0306.2020100152.

基于计算机视觉的茶叶品质在线评价系统

Online Evaluation System of Tea Quality Based on Computer Vision

  • 摘要: 为实现茶叶品质在线评价及自动分级,摒除茶叶品质人工感官审评存在的缺陷,本文研发一套基于计算机视觉技术的茶叶品质在线评价及自动分级系统。采用Open CV和Visual C++开发茶叶品质在线评价软件,结合监督正交局部保持投影方法(supervised orthogonal locality preserving projections,SOLPP)对图像特征变量进行降维处理,对比随机森林(random forest,RF)、反向传播神经网络(back-propagation artificial neural network,BP-ANN)和相关向量机(relevance vector machine,RVM)茶叶品质在线评价模型,得出随机森林算法所建模型性能最好。系统自动完成茶样图像采集、原始图像预处理、特征提取、基于所建模型对待检茶样进行等级评价。控制系统根据评价结果,驱动分级及收集装置将检测茶样输送到相应等级槽中。经测试,研发系统对市售婺源仙芝绿茶、碧螺春绿茶的分级准确率达到93.00%以上。本系统结构简单,运行稳定,能将待检茶样准确送入到相应等级槽中,满足茶叶等级在线评价要求。

     

    Abstract: In order to realize the online evaluation and automatic grading of tea quality, and eliminate the defects of artificial sensory evaluation of tea quality, this paper focused on development ofa set of tea quality online evaluation and automatic grading system basing on computer vision technology. The software was designed by Open CV and Visual C++ realizingthe online evaluation for tea quality, combining supervised orthogonal locality preserving projections (SOLPP) to reduce the dimensionality for image features variables. The sensory evaluation models of tea quality were respectively developed based on RF (random forest), BP-ANN (back-propagation artificial neural network) and SVR (relevance vector machine), by contrast, the performance of based RF model was the optimal. The system automatically completed the image acquisition of tea samples, original image preprocessing, feature extraction, and the grade evaluation to the tea samples based on the developed models. According to the evaluation results, the tea samples were transfered to the corresponding grading tank by the grading and collecting device droved by the control system. The prototype testing results showed that the classification accuracy rate of the Wuyuan Xianzhi green tea and Biluochun green tea reached more than 93%. This developed system had simple structure with stable operation, and the samples could be accurately sent to the corresponding grade tanks, which could meet the requirements of online evaluation of tea grade.

     

/

返回文章
返回