An Improved Histogram-Shifting-Imitated reversible data hiding based on HVS characteristics

Abstract

In this paper, we propose a reversible data hiding scheme to hide a secret message into a cover image by considering the characteristics of Human Visual System (HVS) in order to improve the visual imperceptibility. The human eyes are more sensitive to the changes in the low intensity pixels than the higher intensity ones. Therefore, we divide the intensity levels (0–255) into four groups: the first group contains 0–79 intensity level; second, third, and fourth group contain, respectively, 80–151, 152–215, and 216–255 intensity levels. We further divide first group into segments of size 2 elements, second, third, and fourth group into 3, 4, and 5 elements sized segments, respectively. After constructing the segments, we scan the image in raster order to identify the peak points for each segment, which are used to embed the secret data. The secret data is also divided into the four segments according to the identified peak points per group. The first segment data is converted into base2 representation, second, third and fourth segment secret data into 3, 4, and 5 base representation, respectively. The first segment of secret data is embedded into the peak points belonging to first group, second, third and fourth group secret data is embedded into the peak points of second, third and fourth group, respectively. Thus, our scheme makes least changes into the pixels belonging to the first group which have least intensity values and most to the fourth group pixels which have highest intensity values. Experimentally, our scheme provides better quality stego image and hides more secret data than the other state of the art schemes. We also build a location map for all the peak points to ensure the reversibility of the proposed scheme.