Visual Cryptography for Multiple Secrets - Visual Cryptography and Secret Image Sharing - page 84

3.4.3 Experimental Results
Here, we implement Shyu et al.'s visual multisecret sharing scheme due to its
generality on the abstraction and the superiority on the pixel expansion (over
Feng et al.'s scheme).
We coded the program by using Borland C++ Builder (BCB) in a personal
computer running MS Windows. Since the blocks are in the shape of chords,
we called the embedded functions in BCB such as circle drawing, line drawing,
ood-lling a closed area, and so on, to build the chord-shaped blocks in the
scheme.
Four experiments were designed to explore the feasibility and applicability
of the visual multisecret sharing scheme. Experiment 1 verifies the correctness
of the scheme for x = 3 where the starting position for encoding on the cir-
cle shares are fixed as above-mentioned. Experiment 2 demonstrates that the
scheme can be easily extended in such a way that the starting position for en-
coding can be arbitrarily assigned. This increases the secrecy of the proposed
scheme. Experiment 3 gives the implementation results of the visual 4-secret
sharing scheme. Experiment 4 presents implementation results of encoding
the shares using cylinder (instead of circle) shares.
Experiment 1: Figure 3.20 illustrates the results of a computer imple-
mentation of the proposed scheme for sharing three secret images. Figures
3.20(a){(c) are the three secrets to be shared, namely P 1 , P 2 , and P 3 , respec-
tively. Figures 3.20(d) and (e) show the circle shares A and B encoded by
Algorithm 1, which expose no information about P 1 , P 2 , and P 3 individually.
Figures 3.20(f){(h) reveal the superimposed results of AB, A 120 B, and
A 240 B, which reconstruct P 1 , P 2 , and P 3 in our visual system, respec-
tively. Figure 3.20(i) gives another superimposed result, A 85 B that leaks
no information about any of the three secrets. In fact, any result of A B,
for = 0 ; 120 ; 240 , is merely a seemingly random picture.
Experiment 2: The encoding processes of A and B in the algorithm start
from the 0 position and move on in a clockwise direction (see Figure 3.5).
However, the starting position for encoding in A (or B) can be predefined
arbitrarily.
Figure 3.21 shows the implementation results of using the same example
as in Experiment 1 with a different starting starting position in B; that is, we
encoded B by starting from the 85 position (85 counterclockwise to the 0
position) while we encoded A by starting from the 0 position as mentioned.
The three secret images are the same as those in Figures 3.20(a){(c). Figures
3.21(a) and (b) are the circle shares A 0 and B 0 encoded by Algorithm 1. Figure
3.21(c) shows the result of A 0 B 0 , which reveals nothing about the secrets,
while Figures 3.21(d){(f) display the superimposed results of (A 0 ) 85 B 0 ,
(A 0 ) 205 B 0 and (A 0 ) 325 B 0 that reconstruct P 1 , P 2 , and P 3 , respectively,
in our visual system. Note that both AB (Figure 3.20(f)) and (A 0 ) 85 B 0