Visual Quality-Enhanced Reversible Data Hiding Scheme Based on Interpolation and Variant Sudoku Matrics

Utilizing interpolation techniques (IT) within reversible data hiding (RDH) algorithms presents the advantage of a substantial embedding capacity. Nevertheless, prevalent algorithms often straightforwardly embed confidential data into non-reference pixels without accounting for the relationships between neighboring pixels. This practice results in suboptimal visual quality in the stego images. To rectify this issue, we have introduced the augmentation and modification interpolation (AMI). Additionally, we have devised two variant Sudoku matrices to facilitate distinct data-hiding strategies tailored for different types of reference pixels: Single type and Paired type. By incorporating these innovative designs, the interpolation operation now preserves the pattern-level information of the block in the original image, rather than focusing solely on the pixel level. In assessing the effectiveness of our proposed scheme, we performed experiments utilizing two representative image datasets. The outcomes of these experiments validate that our scheme attains an average PSNR of 44.11 dB and an average hiding capacity of 1.63 bpp. Consequently, it is affirmed that the advantage offered by our proposed scheme is well-suited for scenarios demanding high-quality stego images.