Author Affiliations
Department of Mechanical Engineering, Tianjin College, University of Science and Technology Beijing, Tianjin 301830, Chinashow less
Fig. 1. System block diagram
Fig. 2. Flow chart of defect detection algorithm
Fig. 3. Preprocessing images.(a) Original hole image; (b) mean filtered image of hole; (c) gray close operation image of hole; (d) gray open operation image of hole; (e) original wire drawing; (f) close operation image of wire drawing; (g) original black thing; (h) close operation image of the black thing
Fig. 4. Images of difference image binarization. (a) Hole image; (b) wire drawing; (c) black thing
Fig. 5. Diagram of defect detection results. (a) Close operation of the hole image; (b) open operation of the hole image; (c) close operation of the wire drawing; (d) close operation of the black thing
Fig. 6. Structural factor characteristic diagram
Fig. 7. Feature maps. (a) Defects of small wire drawing; (b) defects of big wire drawing; (c) defects of hole; (d) defects of fold; (e) defects of the black thing
Fig. 8. Diagram of defect detection results
No. | Condition | Classification |
---|
1 | G≥5% and L≥100 | Big wire drawing | 2 | G≥5% and100>L≥30 and S>7 | Small wire drawing | 3 | G<5% | Black stripes | 4 | G≥5% and 100>L≥30 and S≤7 | Black thing | 5 | A≥200 and 0.1≥D≥-0.5 | Hole | 6 | D<-0.5 or D>0.1 | Fold |
|
Table 1. Feature and defect classification matching table
No. | Specie | Sample | Success number | Error number |
---|
1 | Nonconforming product | 400 | 389 | 11 | 2 | Hole defects | 48 | 46 | 2 | 3 | Drawing defects | 52 | 51 | 1 | Total | 500 | 486 | 14 |
|
Table 2. Test classification results