Evolution of Visual Cryptography Basis Matrices with Binary Chromosomes

Abstract

Visual cryptography is a form of secret sharing requiring only human vision to decrypt a secret image, based on the stacking of seemingly random shadow images, or "shares". The main optimization metrics are pixel expansion and recovered image contrast. Maximization of the latter is crucial, as it determines the quality of the decrypted secret. This paper proposes genetic algorithms to evolve basis matrices striving for maximal contrast. They are encoded as binary chromosomes and an objective function is proposed to evaluate their respective fitness values. Evidence for the optimality of GA parameters is demonstrated experimentally, and the success this approach is shown with up to four shares. Furthermore, the algorithm is extended to a generalized graph access for comparison with other research, and the security of this method is analyzed and confirmed.