We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
基于 OTSU 图像分割算法的碎米检测.
- Authors
陈浩然; 范方辉; 牟天
- Abstract
Broken rice, as a common product in the rice processing process, often affects the taste and taste of the product. Therefore, it is particularly important to screen out the broken rice from the head rice. In response to the above issues, a logistic regression model based on the OTSU image segmentation algorithm was established to detect broken rice in the head rice. After comparing the detection results with the national standard method, it showed that the area under the curve (AUC) value of the logistic regression model was 0.987; the Kolmogorov- Smirnov (KS) value was 0.909, and 0.5 was the optimal threshold. The AUC value of the national standard method was 0.922;the KS value was 0.669, and 21 was the optimal threshold. It could be seen that the accuracy, precision, recall, and F1 Score of the logistic regression model established were superior to the national standard method. In addition, the AUC value of the logistic regression model was closer to 1 than that of the national standard method, and the KS value was higher. Therefore, the logistic regression model could better distinguish broken rice from head rice. The four characteristic parameters of long axis (x1), area (x2), short axis (x3), and long- to-short axis ratio (x4) were all significant influencing factors in the model, and the corresponding linear relationship was z=-139.97-5.35x1+10.93x2+2.86x3+34.59x4.
- Subjects
LOGISTIC regression analysis; IMAGE segmentation; RICE processing; RICE products; REGRESSION analysis
- Publication
Food Research & Development, 2023, Vol 44, Issue 20, p175
- ISSN
1005-6521
- Publication type
Article
- DOI
10.12161/j.issn.1005-6521.2023.20.024