Savunma Bilimleri Dergisi

1303-6831 2148-1776

Millî Savunma Üniversitesi

10.17134/khosbd.1901133

Digital Forensics

Dijital Adli Tıp

Çok Aşamalı Dijital Belge İş Akışlarında Klasik Steganografi Yöntemlerinin Karşılaştırmalı Dayanıklılık Analizi

Comparative Durability Analysis of Classical Steganography Methods in Multi-Stage Digital Document Workflows

https://orcid.org/0009-0001-3998-054X

Temiz

Yücel

MİLLİ SAVUNMA ÜNİVERSİTESİ

https://orcid.org/0000-0003-1000-2619

Gürler Ari

Berna

MİLLİ SAVUNMA ÜNİVERSİTESİ

Advanced Online Publication 03 02 2026 04 05 2026

2002

The Journal of Defense Sciences

Bu çalışmada klasik görüntü steganografi yöntemlerinin dijital belge iş akışları altındaki dayanıklılığı sistematik olarak analiz edilmiştir. LSB, DCT ve DWT tabanlı gömme yöntemleri; JPEG sıkıştırma, PDF rasterizasyonu ve yeniden boyutlandırma gibi çok aşamalı dönüşümler altında değerlendirilmiştir. Deneyler Microsoft COCO veri setinden seçilen 100 görüntü üzerinde gerçekleştirilmiştir. Her görüntüye hata düzeltme kodu ile desteklenen sabit uzunlukta mesaj gömülmüş ve dönüşüm sonrası Bit Error Rate (BER), ECC başarı oranı ve Perfect Recovery Rate (PRR) metrikleri hesaplanmıştır. Sonuçlar, yüksek PSNR ve SSIM değerlerinin her zaman başarılı mesaj geri kazanımı anlamına gelmediğini göstermektedir. LSB yöntemi tüm kalite seviyelerinde başarısız olurken, DCT yöntemi yüksek kalite faktöründe başarılı ancak orta seviyede ani performans kaybı göstermiştir. DWT yöntemi ise kademeli bozulma karakteristiği sergileyerek daha kararlı sonuçlar üretmiştir. Bulgular, klasik yöntemlerin modern belge üretim zincirleri altında farklı davranış sergilediğini ortaya koymaktadır.

This study presents a systematic robustness analysis of classical image steganography methods under multi-stage digital document workflows. LSB, DCT, and DWT-based embedding approaches were evaluated under JPEG compression, PDF rasterization, and resizing scenarios. Experiments were conducted on 100 images selected from the Microsoft COCO dataset. A fixed-length message supported by error correction coding was embedded into each image, and post-transformation performance was assessed using Bit Error Rate (BER), ECC success rate, and Perfect Recovery Rate (PRR). Results reveal that high visual quality metrics such as PSNR and SSIM do not necessarily guarantee successful message recovery. The LSB method failed under all compression levels, while the DCT method exhibited a sharp threshold behavior, succeeding only at high-quality factors. In contrast, the DWT-based approach demonstrated gradual degradation characteristics and maintained higher recovery stability at practical compression levels. The findings highlight that classical steganographic techniques behave differently within modern document production chains and emphasize the importance of channel-aware robustness evaluation.

Steganography Digital document workflow JPEG compression Wavelet transform Robustness analysis

Steganografi Dijital belge iş akışı JPEG sıkıştırma Dalgacık dönüşümü Sağlamlık analizi

[1] A. Westfeld, “F5—A Steganographic Algorithm,” 2001, pp. 289–302. doi: 10.1007/3-540-45496-9_21.

[2] I. A. Khalifa, S. M. Saleem, and A. Sengur, “Multilayer Steganalysis in the Encryption Era: A Comprehensive Review,” European Journal of Applied Science, Engineering and Technology, vol. 4, no. 1, pp. 73–99, Dec. 2025, doi: 10.59324/ejaset.2026.4(1).04.

[3] B. Ari, N. Sobahi, Ö. F. Alçin, A. Sengur, and U. R. Acharya, “Accurate detection of autism using Douglas-Peucker algorithm, sparse coding based feature mapping and convolutional neural network techniques with EEG signals,” Comput. Biol. Med., vol. 143, p. 105311, Apr. 2022, doi: 10.1016/j.compbiomed.2022.105311.

[4] J. Fridrich, M. Goljan, and R. Du, “Reliable detection of LSB steganography in color and grayscale images,” in Proceedings of the 2001 workshop on Multimedia and security new challenges - MM&Sec ’01, New York, New York, USA: ACM Press, 2001, p. 27. doi: 10.1145/1232454.1232466.

[5] Stéphane Mallat, A Wavelet Tour of Signal Processing. Elsevier, 2009. doi: 10.1016/B978-0-12-374370-1.X0001-8.

[6] I. Cox, “Digital Watermarking,” J. Electron. Imaging, vol. 11, no. 3, p. 414, Jul. 2002, doi: 10.1117/1.1494075.

[7] B. Ari, K. Siddique, O. F. Alcin, M. Aslan, A. Sengur, and R. M. Mehmood, “Wavelet ELM-AE Based Data Augmentation and Deep Learning for Efficient Emotion Recognition Using EEG Recordings,” IEEE Access, vol. 10, pp. 72171–72181, 2022, doi: 10.1109/ACCESS.2022.3181887.

[8] M. Tancik, B. Mildenhall, and R. Ng, “StegaStamp: Invisible Hyperlinks in Physical Photographs.”

[9] T.-Y. Lin et al., “Microsoft COCO: Common Objects in Context,” 2014, pp. 740–755. doi: 10.1007/978-3-319-10602-1_48.

[10] N. Provos and P. Honeyman, “Hide and seek: an introduction to steganography,” IEEE Secur. Priv., vol. 1, no. 3, pp. 32–44, May 2003, doi: 10.1109/MSECP.2003.1203220.

[11] J. Fridrich, Steganography in Digital Media. Cambridge University Press, 2009. doi: 10.1017/CBO9781139192903.

[12] G. K. Wallace, “The JPEG still picture compression standard,” IEEE Transactions on Consumer Electronics, vol. 38, no. 1, pp. xviii–xxxiv, 1992, doi: 10.1109/30.125072.

[13] B. F. Barni Mauro, Watermarking Systems Engineering. CRC Press, 2004. doi: 10.1201/9780203913512.

[14] C. I. Podilchuk and E. J. Delp, “Digital watermarking: algorithms and applications,” IEEE Signal Processing Magazine, vol. 18, no. 4, pp. 33–46, 2001. doi: 10.1109/79.939835.

[15] A. Arı, “Analysis of EEG signal for seizure detection based on WPT,” Electron. Lett., vol. 56, no. 25, pp. 1381–1383, Dec. 2020, doi: 10.1049/el.2020.2701.

[16] V. K. Goyal, “Multiple description coding: compression meets the network,” IEEE Signal Process. Mag., vol. 18, no. 5, pp. 74–93, 2001, doi: 10.1109/79.952806.

[17] N. Farvardin and V. Vaishampayan, “Optimal quantizer design for noisy channels: An approach to combined source - channel coding,” IEEE Trans. Inf. Theory, vol. 33, no. 6, pp. 827–838, Nov. 1987, doi: 10.1109/TIT.1987.1057373.

[18] P. Brémaud, “Subband Coding,” in Mathematical Principles of Signal Processing, New York, NY: Springer New York, 2002, pp. 115–126. doi: 10.1007/978-1-4757-3669-4_7.

[19] J. Zhu, R. Kaplan, J. Johnson, and L. Fei-Fei, “HiDDeN: Hiding Data With Deep Networks,” 2018, pp. 682–697. doi: 10.1007/978-3-030-01267-0_40.

[20] N. Sobahi, B. Ari, H. Cakar, O. F. Alcin, and A. Sengur, “A New Signal to Image Mapping Procedure and Convolutional Neural Networks for Efficient Schizophrenia Detection in EEG Recordings,” IEEE Sens. J., vol. 22, no. 8, pp. 7913–7919, Apr. 2022, doi: 10.1109/JSEN.2022.3151465.

[21] B. Gurler Ari, “Non-intrusive load monitoring using Short-Time Fourier Transform (STFT)-based Local Phase Quantization (LPQ) features and hybrid deep–classical learning models,” PeerJ Comput. Sci., vol. 12, p. e3717, Mar. 2026, doi: 10.7717/peerj-cs.3717.