Enhanced Image Steganography System based on Discrete Wavelet Transformation and Resilient Back-Propagation

Image hiding is a form of steganography that works by embedding data into a digital media for the purpose of identification, annotation, and copyrighting. This paper introduces a novel image steganography system, which embeds (RGB) secret image within (RGB) cover image chosen by an enhanced resilient back propagation neural network. The proposed system includes embedding and extraction phases. Three main stages are included within the embedding phase, which are; best cover image selection and processing stage, secret image selection and processing stage and best embedding threshold selection stage respectively. Best cover image is performed using SOM and ERBP algorithms. Secret image is processed by separating it into (Red, Green, Blue) color layers and DWT is then applied. The color layers are then converted to bit streams; modified FLFSR in turns will be used to encrypt these streams to get more secure system. ERBP is again used to select the best embedding threshold values. The extraction phase will be performed the stego image result from embedding process. The performance has been evaluated during embedding and extraction stages considering using several cover and secret images and considering several sizes. Experiments show that (PSNR) is improved efficiently.