An efficient multi-predictor reversible data hiding algorithm based on performance evaluation of different prediction schemes

SARDH: A novel sharpening-aware reversible data hiding algorithm

“…The second algorithm we discuss here is the algorithm presented by Jafar et al [22]. This algorithm is used in the proposed algorithm to embed the compressed location map as well as part of the data when the pixels under consideration satisfy some condition as we will discuss in the following section.…”

“…In order to perform embedding in these pixels, any reversible data hiding algorithm that introduces little change in the pixels values can be used. In SARDH algorithm we use the algorithm that is proposed in [22], which utilizes multiple predictors, in order to achieve good embedding capacity in these regions with minimal distortion. However, since the proposed algorithm uses a block size of 2 Â 2 at each pixel, many of the predictors that are employed in [22] are not applicable.…”

“…In SARDH algorithm we use the algorithm that is proposed in [22], which utilizes multiple predictors, in order to achieve good embedding capacity in these regions with minimal distortion. However, since the proposed algorithm uses a block size of 2 Â 2 at each pixel, many of the predictors that are employed in [22] are not applicable. Alternatively, four different and simple prediction values are used in the proposed algorithm.…”

“…Effectively, for pixel X i,j that belongs to C 1 or C 5 , the minimum, maximum, median and average values, B min i;j , B max i;j , B med i;j and B avg i;j , respectively, of the pixels X i+1,j , X i,j+1 and X i+1,j+1 in the current block are used as the prediction values. Afterwards, embedding of data bits proceeds as discussed in [22], where the polarity of the four prediction error values decides on how the data bits are inserted and how the pixels are modified.…”

“…An adjustable prediction-error expansion scheme that is free of the location map was proposed in [21]. In other algorithms, increasing the embedding capacity was approached by proposing algorithms that compute predictions based on multiple predictors [22][23][24][25].…”