A novel Image Steganography System based on Hybrid Artificial Neural network with Haar–Discrete Wavelet Transform

Prof. Reyadh Naoum; Ahmed Shihab; Dr. Sadeq AlHamouz

Volume : V, Issue : III, March - 2015

A novel Image Steganography System based on Hybrid Artificial Neural network with Haar–Discrete Wavelet Transform

Prof. Reyadh Naoum, Ahmed Shihab, Dr. Sadeq Alhamouz,

Abstract :

Steganography was emerged as a common process for hiding data that are needed to be exchanges between two communication parties within several critical applications; especially security ones. A new steganography scheme is introduced in this paper. An (RGB) secret image is embedded into (RGB) cover image. Enhanced resilient back propagation neural network and Self Organizing Mapping were employed to select the best cover image compatible with the secret image. Two main phases are included within the proposed steganography scheme; the embedding and extraction phases. The secret image is firstly separated into (Red, Green, and Blue) color layers; each layer is then processed by applying DWT. The color layers are then converted to bit streams; these bit streams are encrypted using modified FLFSR to increase the security level of the proposed system. The threshold values during the embedding and extraction phases were selected employing ERBP algorithm. The results ensured the effectiveness of the proposed steganography scheme in terms of low MSE and high PSNR in comparison with the steganography systems introduced and implemented in the literature.