IEEE Consumer Electronics Magazine - September 2018 - 43

i = 1, 2, f, n, from n + 1 shared images. The n recovered
images are similar to n cover images and have a high probability of deceiving an attacker. Because the output of the proposed scheme was identical to the output of the conventional
secret-sharing scheme, it stopped the attacker from using various attack methods on shared images to reveal the secret
image. After applying the concept of secret sharing, steganography was used on n recovered images R i, i = 1, 2, f, n
and the stego image ST, to get the recovered secret image. To
generate stego image ST, an additive modulo operation was
used on n + 1 shared images S j, j = 1, 2, f, n, n + 1. After
generation of n + 1 shared images, temporarily recovered images were generated by using the additive modulo
operation on shared images S j, j = 1, 2, f, and n, n + 1,
respectively. Thus, the generalized equation for produ cing temporarily recovered images TR i, is i = 1, 2, f, n;
TR i = S i + 1 - S i, i = 1, 2, f, n. Subsequently, on the temporarily shared images TR i, i = 1, 2, f, n, the additive inverse
modulo was implemented to generate n recovered images R i,
i = 1, 2, f, n. In level 2, cryptography was used for decryption. The additive modulo operation was used on stego image
ST with i times of n recovered images R i, i = 1, 2, f, n to get
recovered secret image G. The proposed decryption technique
diagram is given in Figure 2.

EXPERIMENTAL RESULTS AND ANALYSIS
The proposed scheme worked efficiently on both color and
grayscale images. For experimentation, we used grayscale
images of dimension 512 # 512 pixels. For binary images, we
had to make some changes in the algorithm, such as removing the multiplication and division operator when updating
the modulo value to two. The experimental results of the proposed scheme are illustrated in Figure 3. We used a secret
image SI [Figure 3(a)] and four cover images C 1, C 2, C 3, C 4,
as illustrated in Figure 3(b)-(e). Figure 3(f)-(j) illustrates
five shared images S 1, S 2, S 3, S 4, S 5, respectively. No shares
revealed any information about the secret image or cover
images. Recovered cover images RC 1, RC 2, RC 3, RC 4 are displayed in Figure 3(k)-(n), with Figure 3(o) as stego image
ST. Recovered secret image, G, which is the same as secret
image SI, is shown in Figure 3(p).

SIMILARITY MEASURES
Here, we discuss the general characteristics and efficiency of
the proposed method by checking the analogy using correlation, PSNR, and MSE. The proposed scheme shared a single
secret image using n cover images to generate n + 1 shares
and worked efficiently for grayscale and color images. The
secret image was not revealed when less than n + 1 shared
images were stacked, and the computation time for encryption and decryption was approximately the same. The proposed scheme used an additive modulo operation, which took
less computation time than the Boolean XOR operation. For
the four cover images, the time taken for encryption and
decryption was . 0.125 and 0.127 s, respectively. We use
three similarity measures, correlation, MSE, and PSNR, to

Data security has been an issue
since the beginning of the digital
era, and researchers have considered many ways to approach it.

....

Sn+1

....
-

TR1

TR2

TRn-1

TRn

....

1
Rn-1

....
*

n -1
*
+

G
- = Additive Inverse Modulo
+ = Additive Modulo
* = Multiplication Operator
ST = Stego Image
Si = i th Shared Image
TRi = i th Temporary Recover Image
Ri = i th Recover Image
G = Recoverd Secret Image
FIGURE 2. The proposed decryption technique.

check the similarity between the input and output of both levels. For correlation, the closer the value to +1 or −1, the more
they are related to each other. A value close to 0 represents
that they are not related to each other. In the MSE, the lower
the value, the higher the similarity between the compared
objects. In PSNR, excellent values range from 30 to 50 dB,
september 2018

IEEE Consumer Electronics Magazine

Table of Contents for the Digital Edition of IEEE Consumer Electronics Magazine - September 2018