Research Article
Industrial Printing Image Defect Detection Using Multi-Edge Feature Fusion Algorithm
Table 2
The confusion matrix and evaluation result of defect detection with different fusion sizes.
| Confusion matrix, accuracy, precision, and recall with different fusion sizes | Fusion size | Truth value | Predicted value | Accuracy (%) | Precision | Recall | Pattern | Defect |
| 1 | Pattern | 0 | 135 | 50.00 | 1 | 0 | Defect | 0 | 135 | 50.00% | 1 | 16 | Pattern | 48 | 87 | 67.77 | 1 | 35.55% | Defect | 0 | 135 | 60.81% | 1 | 64 | Pattern | 110 | 25 | 88.88 | 95.65% | 81.48% | Defect | 5 | 130 | 83.87% | 96.29% | 128 | Pattern | 116 | 19 | 91.11 | 95.86% | 85.92% | Defect | 5 | 130 | 87.24% | 96.29% | 160 | Pattern | 127 | 8 | 95.18 | 96.21% | 94.07% | Defect | 5 | 130 | 94.20% | 96.29% | 192 | Pattern | 127 | 8 | 84.81 | 79.37% | 94.07% | Defect | 33 | 102 | 92.72% | 75.55% | 256 | Pattern | 127 | 8 | 81.11 | 74.70% | 94.07% | Defect | 43 | 92 | 92.00% | 68.14% |
|
|