Autoencoder-Based Image Steganography With Least Significant Bit Replacement

Information confidentiality is critical to the progress of sustainable smart cities. Cover images are commonly used in classic image steganography to safely hide/conceal hidden information or images. This paper proposes a new autoencoder (AE) neural network-based image steganography algorithm. With AE, the model employs the Least Significant Bit (LSB) Replacement technique, embedding a secondary image within the LSB of the pixel values in the cover image. During the extraction phase, the decoder uses the learned features to rebuild both the original cover image and the stego secondary image, allowing for the covert integration of a whole image within the AE framework. The model was trained and evaluated using the MNIST and CIFAR-10 datasets. The suggested method’s effectiveness was measured using three performance assessment metrics: mean square error, peak signal-to-noise ratio, and structural similarity index. The best results were $0.0018,47.89$, and 0.96, respectively, which are promising and outperforming the findings of the most significant bits technique. The model was successful in concealing the secret, as seen by several similarities among the cover and stego images.