An Innovative Steganographic Model using Artificial Neural Network and Discrete Wavelet Transform

Abstract

Steganography has emerged as a popular method for hiding data required for interchange within several
critical applications between two communicating parties. This paper presents an innovative method of
steganography technique. In the RGB cover image, a RGB hidden image is embedded. Augmented
adaptable reverse propagation neural network and Self Organizing Feature Mapping (SOFM) is used for
selecting the supreme compatible host image. The suggested steganography involves duplet major
stages: the stages of embedding and extraction. The hidden picture is split into green, blue, and red color
covers. Next, Discrete Wavelet Transform is appertained to all of these layers. The color covers are then
transformed to bit sequences; the bit sequences are converted to cipher streams using updated Fibonacci
Linear Feedback Shift Register (FLFSR), to enhance the proposed system's security level. Threshold
quantities were chosen using the Enhanced Resilient Back Propagation (ERBP) scheme during the
hiding and extraction phases. The outcomes ensured that the system has a less MSE and elevated PSNR
in accordance with the steganography techniques and applied frameworks conveyed in the literature
review