Image Authentication and Restoration using Block-Wise Fragile Watermarking based on k-Medoids Clustering Approach

ABSTRACT This paper proposes a block-wise fragile watermarking scheme based on k-medoids clustering approach. Proposed scheme is effective enough to reconstruct the tampered extensive content of an image. According to the suggested algorithm, first of all image is divided into the blocks and forty eight bits are calculated for each block, which consist of forty five Recovery bits and three Authentication bits namely Union bit, Affiliation bit and Spectrum bit. Authentication bits for a particular block are calculated by extracting five MSBs from each pixel of that block and apply them, on some predefined hash functions. Forty five Recovery bits are calculated using the means of derived clusters and its corresponding mapping bits. These Forty eight bits for each block are pseudorandomly mapped with some other block using secret key. At the receiver end by comparing extracted and further calculated Authentication bits , one can be acquainted with the tampered block and extracted Recovery bits from mapping blocks gives adequate information to recover extensive content of host image.