| 张帆1，李跃凡2，屈国怡3，杨会军3，郑小梅3，李琪2，于修烛2.基于数字图像比色法的食用油碘值检测模型构建[J].中国油脂,2026,52(2):.[HANG Fan1, LI Yuefan2, QU Guoyi3,YANG Huijun3,
ZHENG Xiaomei3, LI Qi2, YU Xiuzhu2.Construction of iodine value detection model for edible oil based on digital image colorimetry[J].China Oils and Fats,2026,52(2):.] |
| 基于数字图像比色法的食用油碘值检测模型构建 |
| Construction of iodine value detection model for edible oil based on digital image colorimetry |
|
投稿时间：2024-10-15 修订日期：2025-09-21 录用日期：2024-11-26
出版日期：2026-02-20
|
| DOI：10.19902/j.cnki.zgyz.1003-7969.240598 |
| 中文关键词: 食用油 碘值 数字图像比色法 颜色参数 |
| 英文关键词:edible oil iodine value digital image colorimetry color parameters |
| 基金项目:陕西省重点研发计划一般项目（2023-YBNY-164）；陕西省科技创新团队项目（2024RS-CXTD-70） |
|
| Author Name | Affiliation | | HANG Fan1, LI Yuefan2, QU Guoyi3,YANG Huijun3,
ZHENG Xiaomei3, LI Qi2, YU Xiuzhu2 | 1.College of Information Engineering, Shaanxi A & F Technology University, Yangling 712100, Shaanxi, China；
2.Engineering Research Center of Grain and Oil Functionalized Processing of Universities of Shaanxi Province,
College of Food Science and Engineering, Northwest A & F University, Yangling 712100, Shaanxi, China
3.Shaanxi Guanzhongyoufang Oil Co. , Ltd. , Baoji 721306, Shaanxi, China |
|
| 摘要点击次数: 1298 |
| 全文下载次数: 439 |
| 中文摘要: |
| 为准确、快速地检测食用油碘值，建立基于数字图像比色法的食用油碘值预测模型。将不同碘值的油脂溶解于环己烷中，并与4 mL韦氏试剂和饱和碘化钾反应而产生不同的颜色，通过智能手机制作的设备采集数字图像，对图像标准化处理、提取关键颜色参数后，通过回归机器学习算法建立颜色参数与韦氏试剂剩余量之间的关系模型，通过由碘值实测值（国标法测定）与韦氏试剂剩余量推导的公式将韦氏试剂剩余量换算成碘值，从而实现食用油碘值的测定。对预测模型进行了验证集验证、盲样验证以及精密度分析。结果显示：经过图像标准化处理后提取了13个与碘值高度相关（R2＞0.75）的颜色参数；6种回归机器学习算法中，线性回归算法的碘值预测效果最佳（均方根误差为0.014 3）；验证集验证、盲样验证和精密度分析结果显示，数字图像比色法和国标法测得的碘值结果具有较高的相关性（r>0.99），进行6次重复试验，数字图像比色法预测碘值的标准差为0.45 g/100 g，低于国标法的0.69 g/100 g，显示出更优的预测精度。综上，基于数字图像比色法构建的食用油碘值检测模型是可行的。 |
| 英文摘要: |
| In order to detect iodine value of edible oil accurately and quickly, an iodine value prediction model for edible oil based on digital image colorimetry was established.The oils with different iodine values were dissolved in cyclohexane and reacted with 4 mL of Wijs reagent and saturated potassium iodide to produce different colors, and then digital images were collected through a device made of a smartphone. The images were standardized and key color parameters were extracted. A relationship model between color parameters and the residual amount of Wijs reagent was established through a regression machine learning algorithm. The residual amount of Wijs reagent was converted into iodine value through a relevant formula derived from the relationship between the measured iodine value(determined by the national standard method) and the residual amount of Wijs reagent, thereby achieving the determination of iodine value in edible oil. The validation set verification, blind samples verification and precision analysis of the model were also conducted.The results showed that 13 color parameters highly correlated with iodine value (R2>0.75) were extracted after image standardization. Among the 6 regression machine learning algorithms, linear regression algorithm had the best prediction effect (RMSE=0.014 3). The results of validation set verification, blind samples verification and precision analysis showed that the iodine values measured by digital image colorimetry and national standard method had high correlation (r>0.99). After 6 repeated experiments, the standard deviation of iodine value predicted by digital image colorimetry was 0.45 g/100 g, which was lower than the 0.69 g/100 g of the national standard method, showing better prediction accuracy.In conclusion, the iodine value detection model of edible oil based on digital image colorimetry is feasible. |
| 查看全文 查看/发表评论 下载PDF阅读器 |
| 关闭 |
|
|
|
