|
| 基于图像纹理分析技术检测食用植物油中动物油的掺伪 |
| Discerning of vegetable oil adulterated by animal fat using image texture analysis technology |
| |
| DOI： |
| 中文关键词: 植物油 掺伪 动物油 胆固醇 纹理分析 |
| 英文关键词:vegetable oil adulteration animal fat cholesterol texture analysis |
| 基金项目: |
|
| Hits: 1212 |
| Download times: 0 |
| 中文摘要: |
| 采用纹理频谱法、分形维数、改进的Hu不变矩和小波共生矩阵技术提取图像纹理特征，通过最小二乘法对图像纹理特征参数和混合油中胆固醇含量进行分析并建立多元线性回归方程。研究结果表明两者之间显著相关，相关系数R2大于0.95。图像纹理分析技术可以检测出掺入猪油含量为8.5%以上的大豆油，该方法计算的胆固醇含量与国标方法（GB/T 15206—1994）测定值之间的最大相对误差为5.75%。说明在误差允许范围内图像纹理分析技术可用于检测植物油掺伪。 |
| 英文摘要: |
| The texture frequency spectrum method，the fractal dimension，the improved Hu moment invariant and wavelet symbiotic matrix technology were adopted to extract image texture features，and the multiple linear regression equation between image texture features parameters and cholesterol content of the adulterated vegetable oil was established by least squares method. The results showed that the correlation between the image texture features parameters and cholesterol content was significant，and the correlation coefficient was above 0.95. The soybean oil adulterated by more than 8.5% of lard could be discerned by the method，and the maximum relative error of cholesterol content between calculated value by the method and measured value by national standard method(GB/T 15206—1994) was 5.75%,which indicated that the adulterated vegetable oil could be discerned by image texture analysis technology within the allowed error range. |
| 查看全文 View/Add Comment Download reader |
| Close |
