基于图像分析的大米透明度定量评价方法  

A Quantitative Evaluation Method of Rice Transparency Based on Image Analysis Technology

作　　者：崔雯雯[1] 周丽聪[1] 黄碧竹 孙永海[1] 

机构地区：[1]吉林大学生物与农业工程学院,长春130022

出　　处：《食品工业》2015年第7期126-129,共4页The Food Industry

基　　金：吉林省应用基础研究项目(201205013);国家自然科学基金项目(31271861);吉林省人才开发资金(基于感官仿生的食品品质评价方法及设备研究与开发)

摘　　要：提出了基于图像分析的大米透明度定量评价方法。设计了大米透明度图像采集设备,采集大米图像,提取透明度特征值,以透明度感官评价值为因变量,以图像透明度特征值为自变量采用逐步引入-剔除法进行多元线性回归分析,建立大米透明度预测模型,并用聚类分析方法对预测结果和透明度感官分类结果进行对比验证。结果表明,该回归方程检验达到了显著性水平,且聚类分析结果与透明度感官分类结果基本一致,正确率达94.12%。基于图像分析的大米透明度定量评价方法是可行的。A quantitative evaluation method of rice transparency was proposed based on image analysis technology.The rice image acquisition equipment was designed, rice image collected, and characteristic value of rice transparency was extracted. Rice lransparency prediction model was established with sensory evaluation value as the dependent variable, the transparency of the image characteristic values as the independent variable, phase in-excluding method was used to multiple linear regression analysis. The results of cluster analysis method and transparency senses classification was compared. Results showed that the regression equation inspection reached the significant level, the classification results basically the same as clustering analysis and the transparency senses, withe accuracy of 94.12%. The evaluation method of rice Wansparency based on image analysis is feasible.

关 键 词：透明度 图像分析 聚类分析 预测模型 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]