Development of low-complexity matrix embedding with an efficient iterative strategy

A novel suboptimal embedding algorithm for binary logos based on a weight approach embedding (WAE) is proposed. An optimal embedding algorithm, developed on the basis of the maximal likelihood algorithm, is aimed at locating the coset leader as an approach to the minimum embedding distortion. By contrast, there is no need to locate the coset leader; instead, a target vector is required. The corresponding weight of the target vector is close to that of the coset leader; the target vector’s weight is discovered in an efficiently iterative manner. In case of a highest operating complexity, in contrast to that of the optimal maximal likelihood (ML) algorithm, the operating complexity of the suboptimal WAE is linearly proportional to the number of code dimensions, because a full search (which would be necessary in an ML algorithm) is not required. The lower embedding efficiency of the proposed algorithm is superior to those of algorithms proposed in previous works, when the logo message length is small.