Scholarlink

<p>With the increasing demand for image security, image encryption has become a key means of protecting sensitive visual information. However, many existing methods uniformly encrypt the entire image, ignoring the unique characteristics of regions of interest (ROIs). Since ROIs often contain critical content, they require stronger protection, while non-ROI (NROI) regions can use lighter encryption strategies. To address this issue, we propose an ROI-focused multi-image encryption algorithm (ROI-MIE). The algorithm enables simultaneous encryption of multiple images, leveraging cross-image correlation to enhance obfuscation and diffusion performance. By introducing an arbitrary segmentation model, it can accurately distinguish between ROIs and NROIs, thus achieving targeted encryption. For ROI encryption, we construct a two-dimensional composite inverse cosine-cosine-sine hyperchaotic system with strong chaos, a wide parameter range, and high randomness to generate a secure keystream. Based on the system, ROI-MIE will perform three rounds of spiral traversal on the RGB channels using different channel orders to achieve pixel-level rearrangement across channels, and then use double-ended queues for chaotic-driven multi-queue pixel reordering. This includes a hybrid enqueue-dequeue phase and a pure dequeue phase, the latter incorporating four diffusion strategies to enhance security. For NROI, the algorithm employs Fisher–Yates scrambling combined with chaotic sequences to achieve lightweight and secure diffusion. This hybrid design significantly improves overall cryptographic performance while ensuring high security.</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

ROI-focused multi-image encryption for multimedia security: a 2D composite arccos–cos–sine system-based scheme

  • 期刊论文

  • 发表日期：

    2026年3月5日

查看全文
  • Chang Liu,
  • Lin Teng,
  • Xiaoying Zhang

摘要

With the increasing demand for image security, image encryption has become a key means of protecting sensitive visual information. However, many existing methods uniformly encrypt the entire image, ignoring the unique characteristics of regions of interest (ROIs). Since ROIs often contain critical content, they require stronger protection, while non-ROI (NROI) regions can use lighter encryption strategies. To address this issue, we propose an ROI-focused multi-image encryption algorithm (ROI-MIE). The algorithm enables simultaneous encryption of multiple images, leveraging cross-image correlation to enhance obfuscation and diffusion performance. By introducing an arbitrary segmentation model, it can accurately distinguish between ROIs and NROIs, thus achieving targeted encryption. For ROI encryption, we construct a two-dimensional composite inverse cosine-cosine-sine hyperchaotic system with strong chaos, a wide parameter range, and high randomness to generate a secure keystream. Based on the system, ROI-MIE will perform three rounds of spiral traversal on the RGB channels using different channel orders to achieve pixel-level rearrangement across channels, and then use double-ended queues for chaotic-driven multi-queue pixel reordering. This includes a hybrid enqueue-dequeue phase and a pure dequeue phase, the latter incorporating four diffusion strategies to enhance security. For NROI, the algorithm employs Fisher–Yates scrambling combined with chaotic sequences to achieve lightweight and secure diffusion. This hybrid design significantly improves overall cryptographic performance while ensuring high security.