基于Contourlet域特征点的医学图像零水印算法

Medical image zero-watermarking algorithm based on Contourlet domain feature points

针对医学图像在传输过程中有可能泄露患者信息的问题,提出一种基于Contourlet域特征点的零水印算法.该算法首先对图像进行Contourlet分解,以此得到低通图像;然后在低通图像中提取出SIFT特征点,并根据特征点的分布自动定位感兴趣区域(ROI);最后筛选出一定数量的特征点,以生成特征向量用于水印嵌入.实验结果表明,该算法在抵抗高斯噪声、JPEG压缩、旋转、缩放、剪切等攻击时具有很好的鲁棒性,可有效保护医学图像中的信息.

Aiming at the problem that patient’s information may be leaked in the process of medical image transmission,a zero-watermarking algorithm based on Contourlet domain feature points was proposed.The algorithm will not affect the quality of the original medical image,and can embed large capacity watermark information.Firstly,Contourlet decomposition is applied to the image to obtain the lowpass image.Then,SIFT feature points are extracted from the lowpass image and the region of interest(ROI)is automatically located according to the distribution of feature points.Finally,a certain number of feature points are screened out to generate feature vectors for watermark embedding.Experimental results show that the algorithm has good robustness against Gaussian noise,JPEG compression,rotation,scaling,shearing and other attacks,so it can effectively protect the information hiding in medical images.