Robust Signature-Based Copyright Protection Scheme Using the Most Significant Gray-Scale Bits of the Image

Table 9

Results and decisions for different approaches of the proposed MSB-based scheme and existing TROI-based [12] and VQ-based [13] schemes by the support vector machines (SVM) [27] and 𝐾-means clustering algorithm [28].

Schemes

Number of miss (%)

To individual attacks (robustness: h = high, m = moderate,

Size of training set

by SVM

by 𝐾-means

Distance^1

l = low, and n = no robustness; and f = uniqueness fail)^2,3

C+

C−

C+

C−

C+

C−

d

0

1

2

3

4

5

6

7

8

9

Over^4

UROI

20

140

0 (0)

0 (0)

0 (0)

0 (0)

30.2

h

h

h

h

h

h

h

m

m

h

h

TBLK

20

140

0 (0)

0 (0)

0 (0)

0 (0)

31.5

m

h

h

h

h

l

l

l

l

m

m

DCTMSB

20

60

0 (0)

0 (0)

0 (0)

0 (0)

31.3

m

h

h

h

h

m

m

l

m

h

h

𝑡-DWT

20

140

1 (5)

0 (0)

7 (35)

0 (0)

25.1

n

h

n

m

h

n

n

n

n

h

n

TROI [12]

20

140

6 (30)

0 (0)

0 (0)

36 (26)

5.1

n, f

l, f

n, f

m, f

l, f

n, f

n, f

n, f

n, f

m, f

n, f

VQ [13]

20

100

11 (55)

0 (0)

6 (30)

0 (0)

9.2

n

m

n

l

h

n

n

n

n

m

n

¹Distance between support vectors.
²Robustness: correctly classified by both SVM and 𝐾-means with h = high PSNR and LRR, m = moderate PSNR and LRR, and l = low PSNR and LRR; no robustness: n = any false negative by SVM or 𝐾-means; and uniqueness fail: f = any false positive by SVM or 𝐾-means.
³Attack numbers 0-9 are referred using Table 3; and
⁴Overall decision.