An Efficient Prediction-and-Shifting Embedding Technique for High Quality Reversible Data Hiding
EURASIP Journal on Advances in Signal Processing
Hindawi Publishing Corporation
An Efficient Prediction-and-Shifting Embedding Technique for High Quality Reversible Data Hiding
Wien Hong 0
Jin-Hua She
0 Department of Information Management, Yu Da University , Miaoli, 361 , Taiwan
The embedding capacity of a histogram-based reversible data hiding technique is primarily determined by the peak height of the histogram. Recently, some studies have tried to embed data in the histogram of prediction errors by modifying the error values and have better embedding efficiency. However, these methods offer no selective embedment mechanism to exclude the positions where the modification in the embedding operation contributes no capacity but merely degrade the image quality. In this paper, a novel coding method for reversible data hiding is presented. A two-stage prediction algorithm that fully exploits the pixel correlations is employed to create more embeddable spaces, and a selective embedment mechanism is used to enhance the image quality. According to the experimental results, the proposed method achieved the highest payload while maintaining the lowest distortion for most standard test images, comparing to other existing histogram-shifting-based reversible data hiding techniques.
1. Introduction
Data hiding is a technique that embeds data into cover
media by slightly modifying their content [
1
] and has been
used in many applications, such as tamper detection [
2
],
copyright protection [
3
], and finger printing [
4
]. When data
are embedded into cover media, the content of the media will
be inevitably modified and thus distortion introduced. The
distortion caused by data embedding is termed embedding
distortion [
5
]. Although the embedding distortion in many
applications is small, the distorted cover media cannot be
recovered to their original state [
6, 7
]. However, some
applications, such as in medical or military usages, allow
no permanent embedding distortion in order to preserve
content fidelity. This demand has highlighted the needs of
reversible data hiding and has drawn much attention in the
recent years [
8–10
].
The reversible data hiding is a technique that allows
extraction of embedded data from the stego media and
exactly restores the marked media to their original states
[
11
]. Many researchers use digital image as the cover
media because they are often transmitted throughout the
Internet, which is easy to be accessed and may arouse a
little suspicious. An image that is used to embed data is
called a cover image, and an image with data embedded
is called a stego image [
12
]. The earliest reversible data
hiding technique reported in the literature is Barton’s
work [
13
]. Afterwards, a number of reversible data
hiding techniques have been proposed to fulfill the insatiate
demands in this field. In 2003, Tian [
14
] proposed a
novel reversible data hiding method with high payload.
In his method, the difference value between paired pixels
is expanded, and a bit can be embedded into the LSB
of the expanded difference. In Tian’s method, n bits can
be embedded into 2n pixels. Alattar [
15
] extended Tian’s
work by increasing the payload without introducing a
noticeable distortion. In Alattar’s method, n bits can be
embedded into n + 1 pixels. Tian and Alattar et al.’s
embedding techniques can be classified as the
expansionembedding technique. For an expansion-embedding
technique, difference values between pixels have to be expanded
to conceal data. Therefore, the embedding distortion is
relatively large. Besides, the selection of embedding position
to avoid the overflow or underflow problem has to pay
the overhead cost, which may significantly reduce the
payload.
In 2006, Ni et al. [
16
] proposed a reversible data hiding
method and achieved very high image quality. They selected
pairs of peak and zero of an image histogram and shift
the histogram bins to leave embeddable spaces for data
embedding. Ni et al.’s embedding method can be classified
as the shifting-embedding technique. In their work, the
maximum payload is limited by the peak height of the cover
image histogram; therefore, the payload is smaller compared
to the expansion-embedding-based reversible data hiding
techniques. In 2007, Thodi and Rodr´ıguez combined the
expansion-embedding and shifting-embedding techniques
and proposed a reversible data hiding method with higher
payload and lower distortion [
5
]. In their works, the
prediction errors are expanded, and data are embedded
into the LSBs of the expanded prediction errors. A better
performance was achieved in Thodi et al.’s method than that
of Tian’s and Ni et al.’s methods.
Recently, some researchers [
17–20
] adopted the concept
of shifting-embedding technique and embedded data into
the prediction error histogram. Since the peak height of the
prediction error histogram is usually higher than that of the
image histogram for most natural images, (...truncated)