Author Affiliations
School of Information Engineering, Guangdong University of Technology, Guangzhou , Guangdong 510006, Chinashow less
Fig. 1. Flow chart of proposed algorithm
Fig. 2. Non-overlapping uniform segmentation
Fig. 3. Similar sub-block extraction process of Soni algorithm
Fig. 4. Similar sub-block extraction process with non-threshold
Fig. 5. Adaptive sub-block combination. (a) Label for each sub-block; (b) sub-block on edge; (c) case of n=6; (d) case of n=8
Fig. 6. Natural similar areas
Fig. 7. Progressive hybrid feature extraction
Fig. 8. Robust test in MICC-F220
Fig. 9. Robust test in MICC-F2000
Fig. 10. Comparison of results on MICC-F220 dataset
Fig. 11. Comparison of results on MICC-F2000 dataset
Parameter | G1 | G2 | G3 | G4 | G5 | G6 | G7 | G8 | G9 | G10 |
---|
/(°) | 0 | 10 | 20 | 30 | 40 | 0 | 0 | 0 | 10 | 20 | Lx | 1 | 1 | 1 | 1 | 1 | 1.2 | 1.3 | 1.4 | 1.2 | 1.4 | Ly | 1 | 1 | 1 | 1 | 1 | 1.2 | 1,3 | 1.4 | 1.2 | 1.4 |
|
Table 1. Type of geometric attack in MICC-F220 dataset
Parameter | G1 | G2 | G3 | G4 | G5 | G6 | G7 | G8 | G9 | G10 | G11 | G12 | G13 | G14 |
---|
/(°) | 0 | 5 | 25 | 70 | 90 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 40 | 30 | Lx | 1 | 1 | 1 | 1 | 1 | 1.2 | 1.5 | 2.0 | 0.7 | 0.5 | 1.4 | 2.6 | 3.4 | 1.4 | Ly | 1 | 1 | 1 | 1 | 1 | 1.2 | 1,5 | 2.0 | 0.7 | 0.5 | 1.7 | 1.3 | 1.2 | 0.7 |
|
Table 2. Type of geometric attack in MICC-F2000 dataset
Dataset | a | b | c | d |
---|
MICC-F220 | G1 | G2,G3,G4,G5 | G6,G7,G8 | G9,G10 | MICC-F2000 | G1 | G2,G3,G4,G5 | G6,G7,G8,G9,G10,G11,G12 | G13,G14 |
|
Table 3. Attack type
Algorithm | TPR /% | FPR /% | F1 /% |
---|
Soni | 97.5 | 8.5 | 94.7 | Non-threshold method | 98.2 | 9.09 | 94.6 | Non-threshold method+hybrid feature | 98.2 | 8.1 | 95.1 |
|
Table 4. Experimental results of each stage
Algorithm | Time /s |
---|
Silva[17] | 39 | Pun[19] | 399 | Pun[27] | 123 | Ryu[6] | 565 | Soni[21] | 38 | Proposed algorithm | 47 |
|
Table 5. Comparison of running time
Algorithm | TPR /% | FPR /% | F1 /% |
---|
CKN | 93 | 11 | 87.2 | Proposed algorithm | 97.2 | 6.4 | 92.9 |
|
Table 6. Performance comparison between proposed algorithm and CKN algorithm