Scholarlink

<p>To address the limited embedding capacity of existing Reversible Data Hiding in Encrypted Images (RDHEI) schemes, this paper proposes a novel method that integrates bidirectional pixel shifting with a dynamic block partitioning strategy. By modeling encrypted image pixels as horizontal and vertical circular queues, the proposed scheme enables dual-layer data embedding through pixel permutation without modifying pixel values, thereby strictly preserving the statistical characteristics of encrypted images. Furthermore, a dynamic block partitioning mechanism is designed to adaptively select between 2 × 2 and 4 × 4 block structures based on block recoverability, achieving an effective balance between embedding capacity and extraction reliability. Extensive experiments conducted on the BOWS-2, BOSSBase, and UCID datasets demonstrate that the proposed method achieves average embedding rates of 0.7156, 0.7241, and 0.6290 bpp, respectively, significantly outperforming state-of-the-art pixel rearrangement–based RDHEI methods. The experimental results also confirm perfect reversibility and strong resistance to statistical analysis.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Reversible data hiding in encrypted images based on horizontal and vertical pixel shifting and dynamic block partitioning

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年3月2日

查看全文
  • Gang Lin,
  • Xianquan Zhang,
  • Chunqiang Yu,
  • Xuemao Zhang

摘要

To address the limited embedding capacity of existing Reversible Data Hiding in Encrypted Images (RDHEI) schemes, this paper proposes a novel method that integrates bidirectional pixel shifting with a dynamic block partitioning strategy. By modeling encrypted image pixels as horizontal and vertical circular queues, the proposed scheme enables dual-layer data embedding through pixel permutation without modifying pixel values, thereby strictly preserving the statistical characteristics of encrypted images. Furthermore, a dynamic block partitioning mechanism is designed to adaptively select between 2 × 2 and 4 × 4 block structures based on block recoverability, achieving an effective balance between embedding capacity and extraction reliability. Extensive experiments conducted on the BOWS-2, BOSSBase, and UCID datasets demonstrate that the proposed method achieves average embedding rates of 0.7156, 0.7241, and 0.6290 bpp, respectively, significantly outperforming state-of-the-art pixel rearrangement–based RDHEI methods. The experimental results also confirm perfect reversibility and strong resistance to statistical analysis.