EEGmark

Transmission of biomedical signals across a network that is both secure and effective while carrying sensitive and vital patient health data is difficult. In this work, an optimized, secure, and robust watermarking technique for Electroencephalogram (EEG) data is proposed. EEG recordings are applied with 3-level lifting wavelet transform (LWT) to divide into sub-bands (SBs). Two-level LWT is utilized to pre-process the grayscale watermark (WM) before embedding. A suitable strength factor is obtained by using the Manta Ray Foraging Optimization (MRFO) technique to make a balance between resilience and visual similarity. A chaotic map is used to encrypt the WM before embedding to provide security to the proposed scheme. Hessenberg decomposition (HD) and singular value decomposition (SVD) are applied on the SB of the host EEG. The performance of the proposed work is assessed in terms of robustness and imperceptibility (IPY). Parameters such as peak signal-to-noise ratio (PSNR), structural similarity measure index (SSIM), Kullback–Leibler Divergence (KL DIV), Jensen–Shannon Divergence (JS DIV), Percentage Residual Difference (PRD), and normalized correlation coefficient (NC) are used to analyze the performance. To analyze the robustness of the suggested scheme, various types of attacks are applied on the watermarked EEG. These attacks include geometric attacks, image processing attacks, filtering attacks, and noise attacks. Simulation results demonstrate that the presented work attains a PSNR value above 58 dB with an SSIM value higher than 0.9998. Comparisons with existing schemes demonstrate that the suggested scheme outperforms existing watermarking techniques.