MultiLayer Reversible Data Hiding Based on Histogram Shifting with High Quality and Capacity

Abstract

A histogram-based reversible data hiding technique is based on shifting the pixels that are between the peak and the minimum points within the whole cover image. Thus the distortion in the marked image caused by the shifted pixels is the same whether embedding large or small watermarks. Also in this technique, the positions of all the pixels that are associated with the selected minimum frequency value have to be recorded in the location map for the reversibility process, even if the positions of these pixels are not found in the embedded part within the cover image. This causes an increase in the overhead information size and conversely reduces the embedding capacity. To solve these problems, a new technique is proposed that considers the size of the watermark and the distribution of the colour values within the cover image to minimise the distortion in the marked image and to improve the embedding capacity within the cover image. The proposed technique determines the optimal pair of the alternatives of peak and minimum points which are named the container and carried colours respectively to represent the watermark within the best sub-image. Therefore, no extra amount of pixels will be modified or saved within the location map. The experimental results show that by applying a reversible data hiding technique based on shifting the histogram by considering the payload size and the distribution of the colour values within the cover image, the marked image quality and the embedding capacity are improved. In comparison to other state-of-the-art works, the proposed method significantly improves the embedding capacity and the marked image quality under different embedding rates. The proposed work can be considered to be both capacity and quality-based.
Copyright © 2015 Praise Worthy Prize - All rights reserved.