Using Deep Learning to Embed Dual Marks With Encryption Through 3-D Chaotic Map

Abstract

Several deep learning (DL) based data hiding techniques have been proposed in recent years. Generally, these techniques embed and extract hidden marks using deep learning networks. Although these methods are imperceptible and robust, many algorithms are being demonstrated to embed a single mark, which may incur the risk of transmission insecurity. To solve these problems, this paper proposes a robust and secure data-hiding method using deep learning for embedding dual marks in the form of images within a carrier image file. This has significant advantages in data tracing, data usage monitoring and multiple property management. Further, the proposed framework uses a chaotic system over the marked file to achieve a high level of security. Extensive experimental results demonstrate that our proposed method using deep learning is superior to the state-of-the-art techniques with significant improvement of 42.33% and 41.62% in terms of visual quality and robustness, respectively.