Extended Visual Cryptography for Photograph Images - Visual Cryptography and Secret Image Sharing - page 106

TABLE 4.1
Comparison of visual cryptography schemes, namely, VSSS, EVCS, and RG,
with photograph images in the case of (2; 2). Here, m denotes the pixel ex-
pansion of each scheme. S and R stand for relative differences of shadow
images and reconstructed secret image, respectively. l (O)
S

and l (O)

R denote the
tone levels of original images to be processed, while l S and l R denote the
tone levels of the resulting (encrypted/decrypted) images. Strictly speaking,
l S and l R should be 2, because every resulting image consists of white and
black subpixels. However, for human visual system, an image can be observed
as a gray-scale image because of halftoning. In this sense, we specify "l S > 2"
for EVCS and "l R > 2" for all three schemes. "mono.," "cont.," "half.," and
"rand." mean a monotone image, continuous-tone image, halftoned binary
image, and random-dot binary image, respectively.

shadow image secret image

scheme m

l (O)
S

l (O)
R

S l S R l R

VSSS 2 (4) 0 1 (mono.) 1 (rand.) 1=2 1 (cont.) >2 (half.)

EVCS 4 1=4 1 (cont.) >2 (half.) 1=4 1 (cont.) >2 (half.)

RG 1 0 1 (mono.) 1 (rand.) 1=2 1 (cont.) >2 (half.)

to be printed on transparencies and reconstructed secret image by stacking
shadow images are binary images converted by halftoning. The pixel expan-
sion of (2; 2) EVCS is m = 4. A tradeo between relative dierences of shadow
image and reconstructed image exists. If we restrict both relative differences
to be the same, the maximum relative differences are S = R =

1

4 .