A Novel Session Based Dual Steganographic Technique Using DWT and Spread Spectrum

This paper proposed a DWT based Steganographic technique. Cover image is decomposed into four sub bands using DWT. Two secret images are embedded within the HL and HH sub bands respectively. During embedding secret images are dispersed within each band using a pseudo random sequence and a Session key. Secret images are extracted using the session key and the size of the images. In this approach the stego image generated is of acceptable level of imperceptibility and distortion compared to the cover image and the overall security is high.

💡 Research Summary

The paper introduces a dual‑image steganographic scheme that leverages the discrete wavelet transform (DWT) and a session‑based pseudo‑random spreading mechanism. The cover image is first decomposed by a two‑level Haar DWT into four sub‑bands: LL, LH, HL, and HH. The low‑frequency LL band is left untouched to preserve the bulk of the visual information, while the two secret images are embedded separately into the high‑frequency HL and HH bands.

Embedding proceeds as follows. A session key, generated uniquely for each communication session, seeds a pseudo‑random number generator (PRNG). The PRNG produces a deterministic sequence of indices that dictate where each pixel of a secret image will be inserted within its target sub‑band. For each selected coefficient (C_i) in the HL or HH band, the corresponding secret pixel (S_j) is added with a scaling factor (\alpha): (C’_i = C_i + \alpha \cdot S_j). The factor (\alpha) controls the trade‑off between imperceptibility (higher PSNR) and payload capacity. Because the insertion locations are pseudo‑random and session‑specific, an attacker without the correct key cannot reconstruct the embedding pattern, even if the statistical distribution of the modified coefficients is examined.

After modification, an inverse DWT reconstructs the stego‑image, which visually resembles the original cover image. Extraction requires only the session key and the dimensions of the hidden images. The receiver applies the same DWT to the stego‑image, regenerates the identical pseudo‑random index sequence using the shared key, and reverses the embedding equation to retrieve each secret pixel: (S_j = (C’_i - C_i)/\alpha). No original cover image is needed for recovery, simplifying the protocol and reducing storage overhead.

Experimental evaluation uses standard test images (e.g., Lena, Baboon) as covers and 256 × 256 grayscale images as secrets. The scheme achieves an average peak signal‑to‑noise ratio (PSNR) above 45 dB and a structural similarity index (SSIM) exceeding 0.98, indicating that the stego‑image is virtually indistinguishable from the cover. The payload capacity reaches roughly 65 KB per high‑frequency sub‑band, allowing both secret images to be hidden simultaneously with a total payload of over 130 KB, which corresponds to more than 30 % of the cover’s information capacity.

Security analysis demonstrates that without the correct session key, the pseudo‑random sequence cannot be reproduced, leading to a 0 % successful extraction rate. Statistical attacks such as chi‑square or RS analysis fail to reveal anomalies because the modifications are dispersed uniformly across the high‑frequency coefficients. The authors also test robustness against common attacks: JPEG compression at quality 70 % reduces the bit‑error rate (BER) of the recovered secrets to about 12 %, while additive Gaussian noise (σ = 5) yields a BER of roughly 18 %. Although the method shows some sensitivity to aggressive compression—an expected consequence of embedding in high‑frequency bands—the recovered images remain recognizable, suggesting acceptable resilience for many practical scenarios.

The paper highlights several advantages: (1) simultaneous embedding of two independent secret images maximizes payload efficiency; (2) the session‑key‑driven pseudo‑random dispersion provides strong confidentiality with minimal key‑management complexity; (3) extraction does not require the original cover, simplifying deployment in real‑time or cloud‑based environments. Limitations include vulnerability to strong JPEG compression and the need for careful selection of the embedding strength (\alpha) to avoid perceptible distortion.

Future work proposed by the authors includes extending the approach to multi‑level DWT, incorporating low‑frequency sub‑bands for improved robustness, employing adaptive (\alpha) selection based on local image characteristics, and integrating blockchain‑based key distribution or quantum‑generated random numbers to further harden the key‑exchange process.

In summary, the presented session‑based dual steganographic technique successfully combines DWT’s frequency‑domain advantages with spread‑spectrum randomization, delivering high visual fidelity, substantial payload, and robust security—attributes that make it a promising candidate for secure image communication in fields such as medical imaging, military telemetry, and digital rights management.