Reversible Data Hiding Using Convolutional Neural Network and Digital Signal Processing Techniques
Peng Qingmei, Li Shaozi, Lin Yaojin, Xiehua Yu · 2022
Reversible data hiding is a data hiding process in which data is encrypted into an image as a means of authentication, and hidden data is used to hide the original image, and RDH is often used in areas where data authenticity and data preservation are important, such as legal/criminal investigations, medical images. In this study, this paper designed a novel reversible data hiding system that utilizes the Steghide algorithm to encrypt images and hide data, as well as an artificial neural network to recover embedded text and structure. Based on the analysis of the test results, it was concluded that the RDH scheme proposed in this article has the same or higher system accuracy and reliability as the existing RDH scheme. To further refine the system results in future studies, this paper proposes to increase the number of training iterations and the number of images in the training dataset to further improve the accuracy and reliability of the development.