A Robust Watermarking Algorithm for Medical Images Based on DWT-Daisy and Discrete Cosine Transform

Abstract

With the development of digital medical imaging technology, the transmission of medical data in the network is becoming more and more frequent. In order to protect patients’ privacy information during transmission, a robust medical watermarking algorithm based on multilevel discrete wavelet transform (M-DWT), Daisy descriptor and discrete cosine transform (DCT) are proposed. The algorithm first performs a three-level DWT transform on the medical image, takes the center point of its low-frequency part LL3 sub-band as the center pixel point of Daisy descriptor, calculates the Daisy descriptor matrix of this point, and performs DCT transformation on it to obtain a 32-bit binary feature vector of the medical image. Then, we associate the feature vector with the watermark to complete the embedding and extraction of the medical watermark. In addition, in order to enhance the security of the watermark information without destroying the medical data information, the watermark is encrypted using the logical mapping, and the zero watermark technology is also adopted. The simulation results show that the proposed algorithm has strong robustness against traditional attacks and geometric attacks. In particular, the algorithm performs better under Gaussian noise attacks and rotation attacks. It also better balances the conflict between the robustness and invisibility of the image watermarking algorithm.