Author Affiliations
School of Electrical Engineering and Automation, Jiangxi University of Science and Technology, Ganzhou, Jiangxi 341000, Chinashow less
Fig. 1. Framework of our algorithm
Fig. 2. Salinas synthetic dataset. (a) False-color image; (b) ground-truth map
Fig. 3. SpecTIR dataset. (a) False-color image; (b) ground-truth map
Fig. 4. HYDICE urban dataset. (a) False-color image; (b) ground-truth map
Fig. 5. San Diego dataset. (a) False-color image; (b) ground-truth map
Fig. 6. Output results of different algorithms. (a) Salinas; (b) SpecTIR; (c) HYDICE; (d) San Diego; (e) ABU-A4; (f) ABU-B3; (g) ABU-U1
Fig. 7. ROC curves of different algorithms on Salinas synthetic dataset. (a) ROC curve; (b) logarithmic ROC curve
Fig. 8. ROC curves of different algorithms on SpecTIR dataset. (a) ROC curve; (b) logarithmic ROC curve
Fig. 9. ROC curves of different algorithms on HYDICE urban dataset. (a) ROC curve; (b) logarithmic ROC curve
Fig. 10. ROC curves of different algorithms on San Diego dataset. (a) ROC curve; (b) logarithmic ROC curve
Fig. 11. ROC curves of different algorithms on ABU-A4 dataset. (a) ROC curve; (b) logarithmic ROC curve
Fig. 12. ROC curves of different algorithms on ABU-B3 dataset. (a) ROC curve; (b) logarithmic ROC curve
Fig. 13. ROC curves of different algorithms on ABU-U1 dataset. (a) ROC curve; (b) logarithmic ROC curve
Parameter | Salinas | SpecTIR | HYDICE | San Diego | ABU-A4 | ABU-B3 | ABU-U1 |
---|
r | 7 | 8 | 6 | 1 | 4 | 5 | 3 | k | 0.7 | 0.8 | 0.65 | 0.45 | 0.7 | 0.7 | 0.1 | q | 0 | 0 | 0 | 0 | 0 | 0 | 0 | Miter | 100 | 100 | 100 | 100 | 100 | 100 | 100 | ε | 0.001 | 0.001 | 0.001 | 0.001 | 0.001 | 0.001 | 0.001 |
|
Table 1. Parameters of our algorithm on different datasets
Algorithm | Salinas | SpecTIR | HYDICE | San Diego | ABU-A4 | ABU-B3 | ABU-U1 |
---|
GRX | 0.9725 | 0.9914 | 0.9857 | 0.8886 | 0.9526 | 0.9998 | 0.9907 | LRX | 0.9232 | 0.9592 | 0.9524 | 0.6246 | 0.7434 | 1.0000 | 0.9674 | LSMAD | 0.9505 | 0.9729 | 0.9914 | 0.9725 | 0.9859 | 0.9996 | 0.9851 | LRASR | 0.9725 | 0.8275 | 0.9959 | 0.8804 | 0.9876 | 0.9998 | 0.8160 | CRD | 0.8273 | 0.9493 | 0.9921 | 0.5605 | 0.5792 | 0.9993 | 0.9803 | Ours | 0.9896 | 0.9995 | 0.9975 | 0.9943 | 0.9974 | 0.9998 | 0.9919 |
|
Table 2. AUC values of different algorithms
Algorithm | Salinas | SpecTIR | HYDICE | San Diego | ABU-A4 | ABU-B3 | ABU-U1 |
---|
GRX | 0.3 | 0.3 | 0.1 | 0.2 | 0.1 | 0.1 | 0.1 | LRX | 60.2 | 52.2 | 24.1 | 44.9 | 40.7 | 44.1 | 45.9 | LSMAD | 24.8 | 29.1 | 9.1 | 15.3 | 15.4 | 15.7 | 16.9 | LRASR | 87.5 | 209.9 | 39.7 | 60.5 | 54.6 | 60.9 | 60.5 | CRD | 27.9 | 65.3 | 14.9 | 19.8 | 19.9 | 19.8 | 20.8 | Ours | 6.4 | 7.5 | 6.7 | 6.9 | 6.8 | 7.1 | 7.0 |
|
Table 3. Detection time of different algorithm unit: s