A Dual-Layer Secure Image Steganography Scheme Using a 6D Hyperchaotic System and Spread Spectrum Modulation with Nearly Lossless Recovery
الباحث الأول:
Ashwaq Auda Kadhim
الباحثين الآخرين:
Doaa F. Al Edhary , Sadiq A. Mehdi, Ali S. Shaker
المجلة:
Ingénierie des Systèmes d’Information
تاريخ النشر:
30 إبريل، 2026
مختصر البحث:
Balancing high embedding capacity, imperceptibility, security, and lossless recovery remains a fundamental challenge in image steganography. This paper proposes a crypto-stego scheme that integrates a six-dimensional (6D) hyperchaotic system with sp…
Balancing high embedding capacity, imperceptibility, security, and lossless recovery remains a fundamental challenge in image steganography. This paper proposes a crypto-stego scheme that integrates a six-dimensional (6D) hyperchaotic system with spread spectrum modulation to achieve dual-layer protection. Unlike existing methods that either sacrifice embedding capacity for security or require complex frequency-domain transforms, our approach embeds a full secret image directly into a cover image of identical dimensions while preserving nearly perfect reconstruction. The proposed scheme generates four distinct keys from a single chaotic sequence, controlling pixel scrambling, spread spectrum modulation, authentication, and final encryption sequentially. Experimental results on BMP, PNG, JPG, and TIFF images of varying sizes demonstrate that the embedded images maintain excellent imperceptibility (PSNR = 43.37–50.78 dB, SSIM = 0.9911–0.9991). Chaotic encryption converts the stego image into random noise (NPCR = 99.60–99.68%, UACI = 33.38–33.47%, entropy ≈ 7.999). The secret image can be recovered with near-perfect quality (PSNR > 81 dB, SSIM = 1.0, correlation = 1.0), and the entire embedding-extraction process completes in under one second. Comparative analysis shows that our scheme outperforms recent deep learning-based steganography methods (U-Net, U-Net++) in both imperceptibility and recovery quality. The scheme also demonstrates graceful degradation under cropping and additive noise attacks. The proposed system offers a practical, computationally efficient solution for applications requiring simultaneous confidentiality, integrity, and real-time performance.
