Lossless embedding using difference between sub-sampled images

Abstract

Kim et al. improved the histogram-shifting embedding technique by exploring the spatial correlation between pixels. The original image had to be partitioned into sub-images. The differences between the sub-images were then calculated for use in hiding data. However after embedding, larger image damage would be impacted by the resulting difference histogram for which an empty bin exists at -1. In this paper, we propose a new shifting method to resolve the problem which results in higher PSNR. Experimental results showed that the proposed method can effectively increase image quality, avoid unnecessary damage and the hidden secret is not easily detected from the stego media.