Convolutional Neural Network Based Data Security in Image Steganography

Abstract

Steganography has made significant progress in recent years but struggles with various obstacles and hurdles. Image steganography refers to the technique of maintaining privacy under a cover picture. This data might take the shape of words, pictures, or videos. This research offers a compact, simple, and fully convolutional design to incorporate a hidden picture within a cover picture and to recover the contained hidden picture from the input image. The paper bases its proposal on an in-depth learning approach and picture-based general steganography techniques. The proposed method uses Convolutional Neural Network (CNN) based steganography to hide the secret information and steganalysis to recover the secret information. Also, it focuses on performance metrics like Peak Signal Noise Ratio (PSNR), Universal Image Quality Index (UQI) and Spatial Correlation Co-efficient (SCC). The outcomes of the experiments have shown that the presented scheme has superior outcomes in terms of concealing capability, confidentiality and resilience and interpretability compared to previous deep-learning picture steganographic techniques.