Detecting Singular Value Decomposition-based WatermarkEmbedded Medical Images using ResNet Neural Network
R. Abirami, C. Malathy · 2023
Digital products copyright protection is a severe issue in the growth of network technology. The security of data, especially medical data, is of utmost importance. The growth of the internet gives easy access to transactional medical data. It causes image distortion and a high risk of misclassification among medical experts. Watermarking aims to protect information and provide authentication for data during transmission. Instead of the conventional method, the deep learning-based watermarking method gained benefits in information security. In this paper, Singular Value Decomposition transformation is applied for embedding watermark, and Resnet a deep-learning model is modified for detecting watermarked images. Watermarking algorithm ensures robust watermarking for medical images to detect changes even after some manipulations like cropping, compression, filtering rotation, and more. MSE, SSIM, and PSNR are some metrics for assessing image quality.