A joint source-channel coding approach to digital image self-recovery

The aim of the digital image self-embedding methods is to restore the content of a tampered image as much as possible. In this article, a joint source-channel coding solution is presented to solve this problem, in which an image compression algorithm is applied to generate a representation of the original image. The preserved information of the original image is used to produce hash bits that help the receiver to find the tampered area. Having the location of tampering known, it can be modeled as an erasure channel. Therefore, a channel coding phase is added to protect the image representation. The limited watermarking bit-budget will be optimally dedicated to the source and channel code bits based on the results derived by solving a dynamic programming optimization problem. It is shown in the experimental results that including this optimization stage results in the superiority of the proposed scheme to the similar state-of-the-art methods who lack such optimization.