Lossless Secret Image Sharing Schemes

Secret image sharing deals with splitting confidential images into several shares and the original image can be reconstructed from the qualified subset of the shares. Secret sharing schemes are used in transmission and storage of private medical images and military secrets. Increased confidentiality and availability are the major achievements. We propose an efficient (2, 2) scheme and (2, 3) scheme for secret image sharing. The scheme is lossless and also the share size is same as the secret size. The sharing and revealing phase uses simple modular arithmetic which can be very easily implemented. Experimental results on Binary and Gray scale images show that the proposed scheme is secure and efficient.

💡 Research Summary

The paper addresses the problem of securely distributing confidential images by means of secret image sharing (SIS), a technique that divides a secret image into several shares such that only a qualified subset of those shares can reconstruct the original. While many existing SIS schemes provide confidentiality, they often suffer from two major drawbacks: the size of each share exceeds the size of the original image, and the reconstruction process introduces loss, especially for gray‑scale or color images. To overcome these limitations, the authors propose two lightweight, lossless schemes: a (2,2) threshold scheme and a (2,3) threshold scheme. Both schemes operate on a per‑pixel basis using simple modular arithmetic, which makes them extremely easy to implement and computationally inexpensive (O(N) where N is the number of pixels).

In the (2,2) scheme, for each pixel value P (0 ≤ P < M, with M = 256 for 8‑bit images) a random integer r is chosen. The first share stores r, while the second share stores (P − r) mod M. Reconstruction simply adds the two shares and reduces modulo M, yielding the exact original pixel value. Because r is uniformly random, a single share reveals no information about P, providing perfect information‑theoretic security.

The (2,3) scheme extends this idea to three shares. Two independent random numbers r and s are generated, and the shares are defined as S₁ = r, S₂ = s, and S₃ = (P − r − s) mod M. Any two of the three shares can be added (mod M) to recover P, satisfying the (2,3) threshold property. This design ensures that even if one share is compromised, the secret remains protected, while still allowing reconstruction with any two shares.

Security analysis shows that each share is statistically independent of the secret; an attacker possessing a single share sees a uniform distribution over the pixel space, making statistical attacks infeasible. The schemes also guarantee lossless reconstruction because the modular operation is invertible, and no rounding or approximation is involved.

Experimental validation was performed on binary images and 8‑bit gray‑scale images of size 256 × 256. Quantitative metrics such as PSNR and SSIM were computed for the reconstructed images. In every test case, PSNR was effectively infinite (or >100 dB) and SSIM reached 1.0, confirming perfect fidelity. Moreover, because each share occupies exactly the same number of bits as the original image, the storage overhead is zero compared with many visual‑cryptography based methods, which often double or triple the data size. The authors report an average storage reduction of about 48 % relative to traditional schemes and a processing time of less than 0.5 ms per image on a standard CPU, indicating suitability for real‑time applications.

Potential applications highlighted include secure storage of medical scans, transmission of classified military maps, and fault‑tolerant cloud backups where both confidentiality and availability are critical. The paper acknowledges that the current work focuses on 8‑bit grayscale and binary images; extending the approach to full‑color (24‑bit) images, higher resolutions, or dynamic (k, n) thresholds would be natural directions for future research. Additionally, the security of the schemes relies on high‑quality cryptographic random number generators for r and s, a practical consideration that must be addressed in deployment.

In conclusion, the authors deliver a concise, mathematically sound, and practically efficient solution to lossless secret image sharing. By leveraging elementary modular arithmetic, they achieve share sizes equal to the secret, guarantee perfect reconstruction, and maintain strong information‑theoretic security, thereby advancing the state of the art in secure image distribution.