Enhancing image quality in Visual Cryptography with colors

The notion of Visual Cryptography was first introduced by Naor and Shamir in 1994 [NS94]. Here “visual” means that the decryption is done by human eyes instead of by any computing devices. To retrieve the ciphertext, simply stack the encrypted images together. In fact, it is more like an image secret sharing scheme than an encryption scheme. Both the secret and the shares are black-and-white pictures, and the shares are drawn on transparent slides to be later stacked together. The shares of the original Naor and Shamir scheme [NS94] are random black-and-white pixels-meaningless images. Chiu [C02] proposed a scheme where shares can be any (meaningful) pictures. Unfortunately, its reconstructed image (stacked share images) is not as clear. To improve the reconstructed image quality, we adopt the color model proposed by De Prisco and De Santis [DD11]. Our input secret image is still black-and-white, but we inject color pixels into the share images. The reconstructed image contains color pixels but looks much more alike to the input black-and-white secret image. The technical difficulty of this work is how and where to inject the color pixels so that both the share and the reconstructed images have high quality.