| 李跃凡1，马改琴1，王媛媛1，张帆2，于修烛1.基于数字图像比色法的食用油酸价检测研究[J].中国油脂,2025,50(7):.[LI Yuefan1, MA Gaiqin1, WANG Yuanyuan1, ZHANG Fan2, YU Xiuzhu1.Determination of acid value of edible oil by digital image colorimetry-based method[J].China Oils and Fats,2025,50(7):.] |
| 基于数字图像比色法的食用油酸价检测研究 |
| Determination of acid value of edible oil by digital image colorimetry-based method |
| |
| DOI：10.19902/j.cnki.zgyz.1003-7969.240191 |
| 中文关键词: 食用油 酸价 图像处理 机器学习 检测 |
| 英文关键词:edible oil acid value image processing machine learning determination |
| 基金项目:陕西省重点研发一般项目（2024NC-YBXM-135）；陕西省科技创新团队项目（2024RS-CXTD-70） |
|
| Author Name | Affiliation | | LI Yuefan1, MA Gaiqin1, WANG Yuanyuan1, ZHANG Fan2, YU Xiuzhu1 | 1.Shaanxi Union Research Center of University and Enterprise for Functional Oil Engineering Technology,
College of Food Science and Engineering, Northwest A & F University, Yangling 712100, Shaanxi,
China
2.College of Information Engineering, Yangling Vocational & Technical
College, Yangling 712100, Shaanxi, China |
|
| 摘要点击次数: 98 |
| 全文下载次数: 55 |
| 中文摘要: |
| 旨在简单、便捷地检测食用油酸价，建立基于数字图像比色法食用油酸价检测方法。通过游离脂肪酸与铜皂染色液反应生成靛蓝色铜皂络合物，用智能手机采集相关的图像，经标准化预处理后，选择合适的颜色参数作为响应值，经回归机器学习建立颜色响应值与酸价模型，对模型进行校准及验证，采用盲样验证方法的准确性，并通过与国标法比较进行精密度分析。结果表明：标准化预处理过程可以减少试验装置、环境及油中色素等显色物质对颜色的影响，提取的15个颜色参数可用于酸价的检测；稳健线性回归算法建立的模型具有最好的预测效果，其均方根误差为0.734 3；与国标法相比，基于数字图像比色法在校准、验证中都显示出较好的预测效果〔相关系数（r）>0.99，标准偏差（SD）<0.25〕，盲样验证的结果也证实了该方法具有较好的准确性和可靠性（r=0.994 2，SD=0.190 2）；精密度分析中基于数字图像比色法的相对标准偏差为4.14%，略小于国标法的（4.16%）。综上，数字图像比色法简单、便捷，可用于食用油酸价的检测。 |
| 英文摘要: |
| Aiming to detect the acid value of edible oil simply and conveniently, a method based on digital image colorimetry for the detection of acid value of edible oil was established. Through the reaction between free fatty acid and copper soap staining solution to generate copper soap complex presenting indigo color, the relevant images were captured with a smartphone, and after standardized preprocessing, the appropriate color parameters were selected as the response values, and the model of the color response values and acid value was established by regression machine learning. The model was calibrated and verified, blind samples were used to verify the accuracy of the method, and precision analysis was carried out by comparing it with that of the national standard method. The results showed that the standardized preprocessing could reduce the influence of the test device, the environment and the color rendering substances such as pigments in the oil on the color, and the extracted 15 color parameters could be used for the detection of acid value. The model established by the robust linear regression algorithm had the best prediction effect with a root mean square error of 0.734 3. Compared with the national standard method, the digital image colorimetry-based method showed better prediction effect in calibration and validation (correlation coefficient (r)>0.99, standard deviation (SD)<0.25), and the results of the blind sample validation also confirmed that the method had better accuracy and reliability (r=0.994 2, SD=0.190 2). The relative standard deviation of the digital image colorimetry-based method in the precision analysis was 4.14%, which was slightly smaller than that of the national standard method (4.16%). In conclusion, the digital image colorimetry-based method is simple and convenient, and can be used for the determination of acid value of edible oil. |
| 查看全文 查看/发表评论 下载PDF阅读器 |
| 关闭 |
|
|
|
