In recent years, histogram shifting based reversible watermarking has become the domain of extensive research due to its high stego-image quality. Most related schemes are required to transmit extra side information, e.g. Peak and zero point pairs, and therefore are non blind in nature. The existing blind schemes, however, exhibit relatively poor performance due to the adoption of fixed peak and zero point pairs. This paper presents a histogram shifting based multi-level embedding scheme for blind reversible watermarking. For single level embedding, a two-step embedding (TSE) approach is proposed, which hides the side information in the stego-image generated with first embedding for blind requirement. While for multi-level embedding, a chain embedding is further incorporated, which ensures both the blind requirement and adoption of optimal peak and zero point pair in each level for high performance reversible watermarking. Extensive simulations are carried out to demonstrate that the proposed scheme could not only achieve the blind reversible watermarking but also outperform other similar ones.%近年来,基于直方图平移的可逆水印算法因其生成水印图像质量高而被广泛研究.目前大部分相关水印算法是非盲的,需要额外传送峰值/零点对等边信息.已有的相关盲水印算法由于采用固定的峰值/零点对而性能较差.本文提出一种面向差值域多层嵌入的高性能直方图平移盲可逆水印算法,有关同步机制可保证在每层嵌入时都选择最优的峰值点和零点.对于单层嵌入,提出两步嵌入法(Two-step embedding,TSE)同步机制,将边信息嵌入到第一步生成的水印像素最低有效位以实现盲提取；对多层嵌入,则进一步提出链式嵌入同步机制,并在最后层采用两步嵌入法以保证水印盲提取.仿真结果表明,本文算法不但可以实现盲提取的要求,而且在算法性能方面较同类算法有显著的提高.
展开▼
