Robust video steganography using wavelet transform with weighted median filter-based prediction model

Abstract

Abstract Several security issues arise when transferring data or information between sites or via public networks. To stop the theft of sensitive data, numerous video steganography techniques have lately been put forth. However, there are numerous problems with these approaches' visual imperceptibility, resilience, and embedding ability. To ensure the robustness of the steganography model, we present robust video steganography employing wavelet transform and a prediction model based on a weighted median filter in this study. Five steps make up the proposed method: frame conversion, discrete wavelet transform (DWT) frame decomposition, secret image encryption, embedding, and extraction. We start by taking into account the two inputs, namely, video and secret information. After that, a frame count is created from the video sequence. We use DWT to divide the frame into low and high-frequency sub-bands after frame conversion. We chose the high-frequency band for the embedding operation from among the sub-bands. Then, we use a secret image to encrypt the data using the Enhanced Cyclic Shift Transposition (ECST) technique. Then, a high-frequency band containing the encrypted secret information is inserted into the cover video. The weighted median-based prediction (WMP) technique is proposed for the embedding process. We use the inverse wavelet transform to obtain stego-information after the embedding process. We use the reverse process while extracting data. The performance of the proposed approach is analyzed based on different metrics and we obtained the maximum PSNR of 57.59dB.