Arabic calligraphy-based text steganography framework with biologically-encoded secret messages, enhanced embedding capacity and automated extraction

Text steganography faces challenges due to limited redundancy in text and the ease of detecting alterations. Linguistic steganography leverages language variability to conceal data, with Arabic offering unique opportunities through its typographical features, particularly in calligraphy. This article introduces a novel framework entitled BioStegoScript that biologically encodes the secret message with dynamic permutations before embedding it into a stego text. Moreover, the proposed framework utilizes the Arabic extension character (Kashida), letter shape variations in the Naskh font alongside white spaces to proficiently embed secret bits. The proposed framework generates stego-texts with improved embedding capacity (≈72.3%), surpassing prior benchmarks of 65.4% and 25.7%. As the stego-text represents its own cover, the security of the proposed framework is robust. Furthermore, the secret message extraction has been automated using Convolutional Neural Networks (CNNs), achieving an accuracy of 98.9%. This automation has simplified the extraction process and enhanced its practicality. These three significant developments establish BioStegoScript as a robust solution for modern Arabic text steganography.