CNN-based Encoder-Decoder Model for Robust and Secure Image Steganography with Enhanced Payload Capacity and Resistance to Steganalysis

Today in the era of fast-growing technology, the issue of secure communication is becoming increasingly important. This is because, with the increase in the volume of data that passes through the internet, a lot of attention has been placed on the security of the information that is being passed. Among the most beneficial strategies to disseminate information securely, steganography is distinguished as the covert transferal of information concealed into other nonsuspect data content like images, audio or video, and so on. While cryptography merely encodes the message there is a good chance the message would be suspected of being coded, steganography is designed to ensure that the covered data is invisible which makes it the most suitable method for secret communication. In this research, a Deep CNN-based encoderdecoder model for image steganography that increases the payload size while also improving its resistance to steganalysis is proposed. Deep learning approaches are employed in the proposed model to provide subtle steganographic marking to the cover images to hide secret information. Experiments carried out show that the proposed model has better payload and detection capabilities than other conventional steganographic techniques. The results prove that the deep learning approach has the ability to revolutionize secure data transmission in digital communications.