Scholarlink

<p>With the increasing demand for image transmission security, the region of interest (ROI) encryption has become a critical task to balance efficiency and security. To accurately encrypt the ROI images, this paper proposes an ROI image encryption algorithm based on YOLOv8 and dynamic Joseph scrambling mechanism. Firstly, the object detection model YOLOv8 is employed to extract the ROI from the image, and the improved non-maximum suppression algorithm can solve the problem that the detection boxes output by the model cannot completely cover the ROI. Secondly, a novel one-dimensional Logistic-Sine chaotic map is proposed, which has excellent chaotic performance. Thirdly, a dynamic Joseph scrambling mechanism is designed to disrupt pixel positions, where the step parameter is adaptively controlled by chaotic sequences. Finally, the mixed-diffusion operations are employed to change the pixel values. The information entropy, NPCR, UACI and encryption time are 7.9964, 99.6059, 33.4978, and 0.2475&#xa0;s on average, respectively. Experimental results demonstrate that the proposed algorithm achieves a superior balance between efficiency and security, and robust resistance against noise and cropping attacks.</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 image encryption algorithm based on YOLOv8 and dynamic Joseph scrambling mechanism

  • 期刊论文

  • 发表日期：

    2026年5月12日

查看全文
  • Xiaoqiang Zhang,
  • Xingwen Sun

摘要

With the increasing demand for image transmission security, the region of interest (ROI) encryption has become a critical task to balance efficiency and security. To accurately encrypt the ROI images, this paper proposes an ROI image encryption algorithm based on YOLOv8 and dynamic Joseph scrambling mechanism. Firstly, the object detection model YOLOv8 is employed to extract the ROI from the image, and the improved non-maximum suppression algorithm can solve the problem that the detection boxes output by the model cannot completely cover the ROI. Secondly, a novel one-dimensional Logistic-Sine chaotic map is proposed, which has excellent chaotic performance. Thirdly, a dynamic Joseph scrambling mechanism is designed to disrupt pixel positions, where the step parameter is adaptively controlled by chaotic sequences. Finally, the mixed-diffusion operations are employed to change the pixel values. The information entropy, NPCR, UACI and encryption time are 7.9964, 99.6059, 33.4978, and 0.2475 s on average, respectively. Experimental results demonstrate that the proposed algorithm achieves a superior balance between efficiency and security, and robust resistance against noise and cropping attacks.