A Randomized Kernel-Based Secret Image Sharing Scheme

This paper proposes a ($k,n$)-threshold secret image sharing scheme that offers flexibility in terms of meeting contrasting demands such as information security and storage efficiency with the help of a randomized kernel (binary matrix) operation. A secret image is split into $n$ shares such that any $k$ or more shares ($k\leq n$) can be used to reconstruct the image. Each share has a size less than or at most equal to the size of the secret image. Security and share sizes are solely determined by the kernel of the scheme. The kernel operation is optimized in terms of the security and computational requirements. The storage overhead of the kernel can further be made independent of its size by efficiently storing it as a sparse matrix. Moreover, the scheme is free from any kind of single point of failure (SPOF).