Complexity-Adaptive Reversible Video Information Hiding Based on Motion Vectors
Yifei Meng, Ke Niu, Yucheng Liang · 2023
In order to address the issue of the existing motion vector-based video reversible information hiding algorithms, which are unable to adaptively adjust the embedding capacity based on the visual characteristics of video frames and have limited capacity, a multi-channel vector sorting reversible video information hiding algorithm is proposed. This algorithm achieves adaptive information embedding for subsequent frames based on the complexity of reference frame texture and motion. Simultaneously, an improvement is made to the multi-pass pixel value ordering (multi-pass PVO) technique and applied to video information embedding, effectively enhancing the embedding capacity of the reversible hiding algorithm. Experimental results demonstrate that compared to similar algorithms, the changes in PSNR and SSIM are reduced by 14.5% and 8.5% respectively, while the embedding capacity is increased by 7.4%. This algorithm exhibits significant improvements in terms of visual quality and embedding capacity.