基于恢复信息共享机制的最优自嵌入算法

提出一种动态的基于恢复信息共享机制的自嵌入算法，可用于实现图像篡改检测和内容恢复. 图像中每个分块产生的水印包括恢复信息和认证信息，其中恢复信息由原图的若干最高有效位生成，认证信息由生成的恢复信息结合最高有效位数据得到. 水印嵌入分为重叠模式和非重叠模式，选取的最高有效位数据层数和生成恢复信息的比特数是动态变化的. 该算法可根据篡改面积选取最优嵌入方式，有效保证了含水印图像的质量和图像篡改恢复效果.实验结果表明：与已有算法相比，该算法可取得更好的篡改恢复性能.

Optimal Self-Embedding Scheme Based on Recovery-Data Sharing

This paper proposes a self-embedding fragile watermarking scheme for image tampering detection and recovery based on variable recovery-data sharing. The generated watermark bits in each image block are composed of recovery bits derived from the interleaved, and scrambled MSB bits of original image and authentication bits derived from the recovery bits and MSB bits. Different from the reported methods using a fixed embedding mode, the proposed scheme has two types of embedding modes, i.e., overlapping embedding and overlapping-free embedding, and uses variable numbers of MSB layers and recovery bits. One can choose an optimal embedding mode to achieve better quality of watermarked image and better performance of content recovery. Experimental results show flexibility and superiority of the scheme as compared with other reported methods.