A New Hybrid Steganography Scheme Employing A Time-Varying Delayed Chaotic Neural Network

Abstract

A secure hybrid audio steganography algorithm using Discrete Wavelet Transform (DWT) and Hopfield Chaotic Neural Network is presented in this paper. An uncompressed audio file is used as a cover medium and a greyscale image is used as secret data. The pixels of the secret image are reordered using cyclic shifting to increase the system security, then the permutated pixels are encoded by applying Hamming code (7,4) before embedding them in the DWT coefficients of the stereo audio signal. The chaotic neural network is applied here to generate a random sequence to choose the embedding locations of hidden image pixels. Regarding the system’s quality, the Peak Signal to Noise Ratio of stego-audio files is above 60 dB, which is close to the original audio quality. Furthermore, the algorithm has an improved embedding payload than previously proposed algorithms and high-security performance, as proved by the results obtained.