Robust Near-Reversible Data Embedding Using Histogram Projection

Most data embedding techniques distort the original image as a side effect. To overcome such a problem, reversible data embedding is used to restore the original content after the removal of the embedded data. However, it is assumed that the embedded image is not altered and such a constraint is sometimes considered unnecessary because as long as the image can be restored to an acceptable level, moderate alteration on the embedded image should be allowed. Such a reasoning leads to a new category of techniques called near-reversible data embedding, in which the embedded image is allowed to be manipulated and the image can be restored to some extent that is very close to the original image. In this paper, a novel near-reversible data embedding algorithm is proposed based on projection histogram manipulations. The experimental results demonstrated that proposed algorithm performs very well in terms of embedding capacity, image fidelity, and robustness against common image processing operations.