Image Secret Sharing - Visual Cryptography and Secret Image Sharing - page 383

compute the image at position 0. In Figure 14.2(b) , we put the shares in x 1 ,
x 2 and obtain the secret color image at 0; and in Figure 14.2(b), we put the
shares at x 1 , x 2 , x 3 and restore the secret image at position 0. For a color
image, this operation can be performed for each color channel. Figure 14.3
shows our experimental results for the Lagrange interpolation scheme of a
color image.
(a) Original image and shares 1 and 2
(b) Original image and shares 1, 2, and 3
FIGURE 14.3
(See color insert.) Experimental results of image sharing based on the La-
grange interpolation in (a) and (b).
In Figure 14.3(a), the original color image is the secret image to be shared,
at least two shares are needed to reconstruct the original image, i.e., it is a
(2;n) scheme, thus k = 2. Thus, share 2 has been generated using the original
color image and share 1 (which is an innocuous image). Now, share 1 and
share 2 can be distributed independently and the original color image can be
reconstructed anytime if we obtain both the shares. In Figure 14.3(b), the
original color image is the secret color image and at least three shares are
required to restore the secret color image. This is because k = 3, therefore
it is a (3;n) secret image sharing scheme. Notice that the secret image is
somewhat visible in the generated shares. We will fix this problem later with
the block-based approach.