Advancement of Digital Image Steganography using Deep Convolutional Neural Networks

Advancement of Digital Image Steganography using Deep Convolutional Neural Networks

연구 분야: Strategies

학회: 2022 International Interdisciplinary Humanitarian Conference for Sustainability (IIHC)

Steganography is a means of secretly concealing information in which the act of steganography itself is hidden in plain sight. Steganography is also known as “stealth writing.” Steganography is not a suitable replacement for cryptography; nevertheless, it can be used in conjunction with cryptography to further strengthen the information hiding mechanism. Steganography is not a good replacement for cryptography. The cover media for a steganographic method can be anything that can be represented digitally, such as a text file, image, audio file, video file, Transmission Control Protocol/Internet Protocol packet, or any other entity. Steganography can be used to hide information in these and other formats. First, the Robust Multitier Spatial Domain Secured Color Image Steganography in Server Environment (MSS-SE) approach is created as part of an effort to lower the possibility of compromising confidential communications. MSS-SE steganography methods, rather than transmitting the original picture from the server, send a randomized mix of a cover image and a confused image from the same server. This is done in place of delivering the original image. A unique approach that is based on deep hierarchical spectral spatial feature fusion has been developed by us in order to improve the classification of HSIs (DHSSFF). The pooling method, which is considered by CNN to be one of the most important regularization procedures, has been the primary focus of our attention prior to the construction of the model. Following an examination of the relevant published material, we have reached the conclusion that a wide variety of pooling algorithms have been applied to carry out fruitful analyses of remote sensing data.