Метод адаптивного LSB-вбудовування на основі аналізу локальної ентропії цифрового зображення

• Datum: 2026
• 1. Verfasser: Слободянюк, Олександр
• Format: Artikel
• Sprache: Ukrainisch
• Veröffentlicht: Kamianets-Podilskyi National Ivan Ohiienko University 2026
• Online Zugang: https://mcm-tech.kpnu.edu.ua/article/view/356387
• Назва журналу: Mathematical and computer modelling. Series: Technical sciences

The article proposes an adaptive LSB steganography method for audio containers based on block Shannon entropy. The method allows selecting for embedding only those frames of the audio signal that are characterized by high local entropy, i.e., areas with maximum statistical uncertainty and textural complexity. The algorithm involves dividing the audio into frames of 256-1024 samples, quantizing the samples to 256 levels, calculating the Shannon entropy for each block, and selecting frames whose entropy exceeds a predefined threshold. Secret data is embedded by replacing the least significant bit (LSB) of samples exclusively in the selected frames. To enhance security, the order of frames can be permuted using a secret key. Information extraction is performed deterministically by repeating the frame selection procedure on the receiver side. Experimental comparison with classical LSB, chaotic LSB (based on chaotic maps), and amplitude-adaptive LSB showed significant advantages of the proposed method. It provides the highest stego-audio quality values: SNR – 52,3 dB and PESQ – 4,7. The probability of detection by steganalysis decreased to 38% compared to 92% for the classical method. Although the hiding capacity is 0.5 bit per sample, the overall trade-off between imperceptibility, robustness, and quality is significantly better than in the compared methods. The theoretical basis of the work lies in the application of an information-theoretic criterion for adaptive data hiding. It is shown that Shannon entropy is a more effective criterion for selecting embedding positions than signal amplitude or pseudorandom sequences, since it better reflects the real statistical complexity of the audio signal and allows natural masking of LSB changes. The proposed method is suitable for covert transmission of confidential information in VoIP systems, audio messengers, podcasts, and streaming services, as well as for hidden watermarking of audio content. The research results confirm the prospects of using block entropy as an adaptivity criterion in audio steganography and fill a gap in the existing literature, where similar methods are widely used for images but are almost not studied for audio containers.