Image Steganography with CNN Based Encoder - Decoder

Abstract

Image steganography is a significant area of research that aims at finding any hidden images or information within other digital images. In this paper, we propose a novel approach combining the Convolutional Neural Network (CNN) model and an encoder and decoder network to ensure accurate and efficient image steganalysis. The proposed method leverages the discriminative power of CNNs to extract features of importance from images and then uses the encoder-decoder networks to reconstruct the original image from the stego or hidden content. Our customized CNN model is designed to capture the features present in the steganographic images. The Encoder-Decoder network plays a crucial role in steganalysis by reconstructing the original image from the hidden content. By training the network on a diverse set of steganographic images, it learns to identify the distinctive artifacts introduced during the steganographic embedding process. The reconstructed image is then compared with the original image using appropriate similarity measures, allowing us to accurately detect the presence of hidden information. The combination of the customized CNN model and the Encoder-Decoder network enables efficient and robust image steganalysis, making it a valuable tool for digital forensics and security applications.