Multi-DS Strategy for Source Camera Identification in Few-Shot Sample Data Sets

Table 7

Average confusion matrix obtained by SVM classification over 20 iterations (%) (Dresden database).

	C1	C2	F1	K1	N1	N2	N3	O1	P1	P2	R1	SL1	SN1	SD1	SD2	SD3
C1	97.5	0.7	0.2	—	0.2	0.1	—	0.3	0.1	0.3	0.1	0.5	—	—	—	—
C2	2.2	90.9	0.1	0.1	0.7	—	—	—	0.6	0.8	—	3.6	—	0.3	0.5	0.2
F1	0.1	—	86.3	0.1	—	0.4	0.3	0.1	1.8	1.5	8.2	0.5	0.5	—	—	0.1
K1	—	—	0.3	94.6	—	0.7	3.6	—	—	0.7	—	—	0.1	—	—	—
N1	0.7	2.3	—	0.7	92.2	1.3	0.8	0.2	—	0.4	—	1.4	—	—	—	—
N2	0.5	0.7	—	1.1	0.7	91.9	3.3	0.1	—	0.3	0.4	0.7	0.3	—	—	—
N3	—	—	0.5	0.8	—	2.9	95.0	—	0.1	1.0	0.2	—	—	0.1	0.3	—
O1	2.2	—	0.3	0.2	0.1	0.5	0.5	93.5	0.1	0.3	0.1	2.1	0.1	—	—	—
P1	0.1	0.9	0.1	0.1	—	0.1	2.0	0.1	93.7	0.2	0.1	2.2	0.2	—	0.1	0.1
P2	0.4	1.3	0.9	1.4	2.1	2.6	—	—	0.8	87.2	1.9	—	1.2	0.1	—	0.1
R1	—	0.1	3.7	0.1	0.5	2.2	0.1	0.1	—	1.0	90.4	0.4	1.4	—	—	—
SL1	1.0	2.1	0.1	—	0.9	1.3	0.2	0.3	0.8	0.4	0.5	92.0	0.4	—	—	—
SN1	—	0.9	1.4	0.8	0.1	0.6	0.1	0.1	0.7	4.3	2.4	0.7	87.4	0.5	—	—
SD1	0.1	0.1	0.3	0.1	—	—	0.9	—	0.2	0.3	0.4	—	0.2	66.5	3.7	27.2
SD2	—	—	0.2	—	—	—	0.7	—	—	0.1	0.4	0.2	—	2.0	93.9	2.5
SD3	—	0.1	0.1	—	—	—	0.7	—	—	0.1	0.1	—	—	50.2	1.3	47.4

The bold values (diagonal values) indicate that the prediction labels of the model are equal to the real labels, that is, the accuracy of the correctly classified results.