V3I8P13

CipherVeil: A Secure Multimedia Steganography Framework with AES-256 and Pseudo-Random LSB Embedding

Mrs. Saranya R^1*, Kaustubh Vishal Mundra², Kush Sandipkumar Patel³

Abstract

Covert communication based on the lossless multimedia carriers has been subject to extensive research due to the possibility of embedding encrypted data payloads, although many of the proposed LSB algorithms are designed deterministically due to non-random placement of the embedded bits. CipherVeil improves the LSB steganography state-of-the-art by implementing deterministic embedding process in both image and video-based carriers via using the same passphrase driven pipeline during extraction to retrace the embedding pattern and thus recover the encrypted data stream. The 32-bit header section is followed by zlib compressed payload protected by the AES 256-CBC encryption routine. The secret key for AES encryption is derived from the provided passphrase via SHA-256 hashing, whereas the second digest is used to determine the pseudo random embedding pattern with NumPy RandomState utilized to populate pseudo-random positions within the RGB channels as opposed to conventional sequential iteration. The lossless PNG and FFV1 MKV codecs were used as the carriers due to their general applicability and resilience against destructive recompression. The highest PSNR of 79.85 dB was obtained for the 512 × 512 pixels PNG image with 18.1 percent of theoretical maximum embedding capacity, which is achieved at 6.72×10−4 MSE. The PSNR values were increasing up to 86.04 dB and 88.98 dB for the 1024 × 1024 and Full HD images at lower embedding densities. Meanwhile, the SSIM index remained practically at 1.0000 throughout the experiments for all three image sizes. The paper reports the deterministic nature of the proposed approach as well as its practical implementation and further research directions such as chi-square, RS, and CNN based steganalysis. The utilization of compression, encryption, randomization, and deterministic extraction with lossless carriers enables such a promising steganographic framework to offer bet ter protection of the embedded data and improved performance as compared to the state-of-the-art LSB methods. Furthermore, due to the deterministic and symmetrical nature of the CipherVeil algorithm, it becomes possible to extract the embedded data with the correct passphrase without additional requirements for synchronization information. The experiments demonstrated the possibility to preserve the carrier’s imperceptibility while distributing the changes across all three-color channels in the image.

Keywords:

Multimedia steganography, AES-256, pseudo random embedding, LSB substitution, passphrase-driven extrac tion, lossless video steganography