Image Steganography-Embedding Secret Data in Images Using Convolutional Neural Networks

Abstract

Image steganography methods use manual features for hiding payload data in cover images. These manual features allow less payload capacity and also cause image distortion. In this chapter, the authors detail a CNN-based network image steganography. The major contributions are twofold. First, they presented a CNN-based encoder-decoder architecture for hiding image. Secondly, they introduce a loss function, which checks joint end-to-end encoder-decoder networks. They evaluate this architecture on publicly available datasets CIFAR10. The results indicate an increase in payload capacity with high peak signal-to-noise ratio and structural similarity index values.