A novel message embedding algorithm using the optimal weighted modulus.

This paper proposes a novel message embedding algorithm using the optimal weighted modulus. Our algorithm, referred to as a w eighted m odulus and abbreviated by WM( n, M ), conceals an M -ary secret digit in a group of n pixels. Using an n -tuple optimal weight ensures that our algorithm can produce stego pixels with minimal pixel distortion. We adopt four steps to pragmatically accomplish the goal of message embedding. In the first step, we introduce a check-weight algorithm to systematically generate all of the valid weights. Secondly, we classify valid weights by referring to theoretically minimal distortion to produce an optimal weight. Along with the classification, we construct a pixel alteration table which simplifies the process of the message concealment through convenient vector operations. We take advantage of the homogeneous alteration table that is built in this step to decrease the distortion encountered when resolving the problem of pixel underflow or overflow. Finally, we propose a universal bit conversion scheme which transforms a serial secret bit stream into M -ary digits to minimize the transformation loss, thereby achieving maximal embedding capacity. Our algorithm accurately predicts results prior to real message concealment, effectively conveys maximal secret bits and speedily produces stego images with theoretically minimal distortion. The experimental results demonstrate that our algorithm outperforms 11 current state-of-the-art competitors. It is not detectable by the SPAM and RS steganalytic attacks. [ABSTRACT FROM AUTHOR]