Reversible data hiding scheme based on improved interpolation and three-in-one intelligent strategy

Abstract

Reversible data hiding method combined with interpolation is a widely used technique in secretly transmitting large collections of information through images. However, this approach can be subjected to inaccurate prediction because of using a single prediction formula for different textures of complex images. We proposed a new method using a circular predictor interpolation (CPI) for improved accuracy performance, which adjusts the predictive radius adaptively for different image textures and takes the coefficient of distance into consideration to increase image load capacity and visual quality. Additionally, to reduce unnecessary image distortion under low load conditions, the first part of the intelligent strategy modifies the scanning sequence in images so that the secret data is preferentially embedded into the pixel with less deviation. Meanwhile, an interpolation-embedding synchronization was created as the second part to allow the interpolation domain and embedding domain to be adjusted synchronously in light of the data capacity to be embedded. In addition, the third part of the intelligent strategy is to take the original image as a benchmark for steganographic pixel value rectification to improve the overall image quality and the overflow or underflow problems could be solved in this process. The outcomes of the experiments indicate that the CPI and the intelligent strategy realize a high payload and play an important role in improving visual quality compared with six advanced schemes.