A high-capacity Coverless Image Steganography Based on Rule based Machine Learning

Abstract

Abstract Securing private information during its transmission over the internet is of paramount importance, as it safeguards against unauthorized access and guarantees the integrity of data. Steganography serves as a technique for enhancing security by concealing sensitive information within digital content such as images, videos, audio, and textual carriers, thus thwarting unauthorized interception. Current methodologies involve manipulating pixels to create stego-images, effectively embedding confidential messages within the image structure. However, this approach has led to steganalysis experts uncovering these hidden messages. To address this issue, an innovative strategy for covert data concealment is proposed. This method involves integrating the secret message through the establishment of a cover file or secret message mapping. This approach ensures the confidentiality of sensitive data communicated over the internet. This research introduces a novel and highly-capable coverless image steganography technique utilizing RBML-based optical mark recognition. Notably, this approach surpasses previous alternatives in terms of embedding capacity, as substantiated by result analysis. Furthermore, it is adaptable for refinement as needed. Significantly, it exhibits remarkable resilience, safeguarding against a spectrum of attacks including scaling, color space conversion, JPEG compression, thresholding, "salt and pepper" noise, file format conversion, and steganalysis tools, among others.