用于CT和MRI图像的轮廓波变换双水印算法-《中国医学物理学杂志》

文章信息/Info

Title:: Double watermarking algorithm for CT and MRI images based on contourlet transform

作者:: 田振宇1; 徐皓1; 崔欣欣1; 刘晶1; 杨建兰2; 1.甘肃中医药大学信息工程学院，甘肃兰州 730000； 2.泉州市正骨医院，福建泉州 362000

Author(s):: TIAN Zhenyu1; XU Hao1; CUI Xinxin1; LIU Jing1; YANG Jianlan2; 1. School of Information Engineering, Gansu University of Chinese Medicine, Lanzhou 730000, China 2. Quanzhou Orthopedic-Traumatological Hospital, Quanzhou 362000, China

Keywords:: double watermarking spot detection medical image watermarking tamper detection

摘要:: 目的：为解决CT与MRI图像版权侵权与恶意破坏问题，通过结合轮廓波变换与斑点检测技术，提出一种基于轮廓波变换自动选择感兴趣区域（ROI）的双水印算法。方法：该算法通过斑点检测与奇异值分解，选取医学图像轮廓波低频域与诊断最相关的位置作为ROI，在ROI生成零水印图像用于保护图像的完整性，并将ROI生成的零水印与版权水印共同嵌入图像的非感兴趣区域。结果：该双水印算法与其他双水印算法相比，载体医学图像遭受攻击后的不可见性有很大提升，结构相似度达到94%以上。结论：斑点检测轮廓波双水印算法对常规攻击与几何攻击具有较好的鲁棒性，且对于载体医学图像视觉质量有明显提升。

Abstract:: Objective To propose a double watermarking algorithm based on contourlet transform and speckle detection to automatically select the region of interest (ROI) for solving the problems of copyright infringement and malicious destruction of CT and MRI images. Methods The method used speckle detection and singular value decomposition to select the most relevant location in the low frequency domain of the medical image contourlet wave as the ROI, generated a zero-watermarked image in ROI for protecting image integrity, and embed the ROI zero-watermark and the copyright watermark together in the region of non-interest. Results Compared with other double watermarking algorithms, the double watermarking algorithm could greatly improve the invisibility of the carrier image after the attack. The double watermarked images had a structure similarity index measure higher than 94%. Conclusion The proposed double watermarking algorithm incorporating speckle detection and contourlet transform has better robustness against conventional attacks and geometric attacks, and significantly improves the visual quality of carrier images.