Using Artificial Neural Network to Analyze Stego Images
Authors: Nkuna MC, Esenogho E, Heymann R
DOI Info: http://doi.org/10.5281/zenodo.5805077
This paper proposed a discrete cosine transform least significant Bit-2 steganography encryption method for embedding the secret data in the cover image. The method overcomes physical signs of pixel modifications while achieving a high data payload. This technique enables data to be hidden in a cover image, while the image recognition artificial neural network checks the presence of any visible alterations on the stego-image. The traditional least significant bit (LSB) and the proposed discrete cosine transform least significant bit-2 (DCT LSB-2) methods were tested for embedding efficiency. The stego-images obtained from the embedding process using the traditional LSB and the proposed DCT LSB-2 encoding algorithms were analyzed using a neural network. Results obtained from the proposed DCT LSB-2 method achieved high data payload and simultaneously minimized visible alterations, and maintained the efficiency of the neural network compared with the traditional LSB. The proposed method has shown an improved stego-system compared to traditional LSB techniques.
Affiliations: Centre for Collaborative Digital Networks, Department of Electrical and Electronic Engineering Science, Faculty of Engineering and Built-Environment, University of Johannesburg, P.O Box 524, Auckland Park, 2006, South Africa.
Keywords: Least Significant Bits, Stego-image, Artificial Neural Network, Artificial Intelligence, Steganography, Discrete Cosine Transfor
Published date: 2021/12/30