Multiple description coding for image data hiding jointly in the spatial and DCT domains

Abstract

In this paper, we propose a new method for hiding a signature image in the host image. We encode the signature image by a balanced two-description subband coder and embed the descriptions in the different portions of the host image. We split the host image into two images from its even and odd rows, and embed the information of one signature description in the first portion of the host image in the spatial domain, and the other description in the second portion in the DCT domain. In both cases, we employ proper masking operation to reduce visibility of embedded information in the host image. At the receiver, the multiple description decoder combines the information of each description to reconstruct the original signature image. We experiment the proposed scheme for embedding gray-scale signature images of 128x128 pixels in the gray-scale host image of 512x512 pixels, and evaluate the system robustness to various attacks.