New Deep Steganography Model Based on Josephus Confusion

New Deep Steganography Model Based on Josephus Confusion

연구 분야: Strategies

학회: International Conference on Intelligent Systems and Advanced Computing Sciences

With the rapid advancements in computers and the internet, digital media has emerged as a preferred medium for concealing information. This paper explores the field of steganography, focusing specifically on image steganography due to the widespread availability and abundance of data within images. The process of image steganography involves two key components: hiding a secret message within a cover image and extracting it from the image. Traditional techniques operate in either the spatial or frequency domain, directly modifying pixel intensities or embedding information in wavelets, respectively. In contrast, deep steganography utilizes neural networks to generate images containing hidden information, leveraging the power of machine learning for more robust concealment and automated learning processes. Further research is required to fully explore its implications in terms of security, detection resistance, and efficiency. Existing work in deep steganography demonstrates promise in terms of information security, as it preserves the visual and statistical properties of the cover image while offering potential resistance to detection. To enhance this work, we propose the addition of a Josephus confusion layer with a novel architecture, presenting promising and satisfactory results.