优化四阈值的直方图对图像可逆数据隐藏方法
61 浏览量
更新于2024-08-27
收藏 2.21MB PDF 举报
本文探讨了一种基于直方图对的可逆数据隐藏方法，该方法通过搜索最优的四个阈值来实现。论文发表在《信息安全与应用》杂志(39卷，2018年)，该期刊由Elsevier出版，可以在其官方网站www.elsevier.com/locate/jisa上获取最新内容列表。作者Guorong Xuan、Xiaolong Li和Yun-Qing Shi分别来自同济大学计算机科学系、北京交通大学信息科学研究所和新泽西理工学院电气与计算机工程系。
传统的可逆数据隐藏（Reversible Data Hiding, RDH）技术通常涉及对掩蔽图像的预测误差进行处理，以嵌入秘密信息。然而，大多数现有算法采用的方法是将预测误差x修改为2x+，这种操作可能导致数据隐藏的不精确性和潜在的信息损失。该研究提出了一种创新策略，即利用直方图对，这是一种更为精细且可能提高数据隐藏效率和安全性的方式。
该方案的核心在于，通过对原始图像的直方图分析，构建特定的直方图对，然后选择四个最优阈值来分割这些直方图。这样做的目的是为了减少操作对原始图像统计特性的影响，同时确保数据嵌入过程的可逆性，即在提取隐藏信息后，原始图像可以恢复到原始状态，不会产生明显的失真。所谓的“双参数平面”可能是指通过这两个阈值参数来定义数据隐藏和提取的区域，它们在设计中起到关键作用。
论文的主要贡献可能包括：
1. **直方图对的选择策略**：作者可能设计了一种算法或方法，用于找到最适宜的数据嵌入和提取的直方图对，这有助于提高数据隐藏的容量和透明度。
2. **优化阈值搜索**：通过优化阈值设置，该方案可能提高了数据隐藏的效率和安全性，避免了传统方法中的过度修改预测误差。
3. **双参数平面理论**：论文可能详细解释了如何利用这两个参数来构建和管理数据隐藏的双重参数空间，从而实现高效和可逆的数据隐藏。
4. **实验评估与性能比较**：文中可能包含对所提方法与现有RDH算法在数据隐藏容量、图像质量和鲁棒性方面的定量和定性分析，以展示其优越性。
这篇研究论文深入探讨了直方图对在可逆数据隐藏中的应用，并通过寻找最佳阈值来优化数据隐藏过程。其成果可能为图像数据保护提供了一种新的解决方案，适用于需要高安全性和可逆性的应用场景。
Journal of Information Security and Applications 39 (2018) 58–67
Contents lists available at ScienceDirect
Journal of Information Security and Applications
journal homepage: www.elsevier.com/locate/jisa
Histogram-pair based reversible data hiding via searching for optimal
four thresholds
Guorong Xuan
a , ∗
, Xiaolong Li
b
, Yun-Qing Shi
c
a
Department of Computer Science, Tongji University, 1239 Siping Rd., Shanghai 20 0 092, China
b
Institute of Information Science, Beijing Jiaotong University, Beijing 10 0 044, China
c
Department of Electrical and Computer Engineering, New Jersey Institute of Technology, 323 M. L. King Blvd. Newark, NJ 07102, USA
a r t i c l e i n f o
Article history:
Available online 21 February 2018
Keywords:
Reversible data hiding (RDH)
Histogram-pair based RDH
Optimal four thresholds
Double parameter-planes
a b s t r a c t
A histogram-pair based image reversible data hiding (RDH) scheme is presented in this paper. It em-
beds data into the prediction-errors generated from a cover image. In doing so, instead of changing a
prediction-error x to 2 x + b as done in most existing RDH algorithms, where b is a to-be-embedded bi-
nary bit, we modify the quantity x to x + b so as to reduce the embedding distortion. Moreover, four
thresholds, i.e., the embedding threshold, the fluctuation threshold, the right- and the left-end histogram
shrinking thresholds, are employed and adjusted to achieve the highest PSNR at a given embedding rate.
The experimental results have demonstrated that, comparing with the state-of-the-arts works, superior
performance can be achieved by the proposed method. Particularly, by the proposed histogram shrinking,
better embedding result can be achieved for the cover image with high peak bins at both sides of the
image spatial histogram (e.g., the JPEG20 0 0 test image Woman).
©2018 Elsevier Ltd. All rights reserved.
1.
Introduction
In the past two decades or so, image reversible data hiding
(RDH) has become an active research field. In the beginning of
this century, the first a few RDH schemes based on modula-256
technique were reported [1] , which can only embed a very small
amount of data for the purpose of image authentication. How-
ever, although works, it suffers from salt-and-pepper noise [2] and
the limited embedding capacity is not suitable for many other ap-
plications. Later, the least significant bit plane of a cover image
[3] or the integer wavelet transform coefficients of a cover im-
age [4,5] are compressed for RDH. But, the embedding capacity is
still not high enough. To handle this issue, a scheme reported in
[6] has largely boosted the payload of RDH in which one bit can
be possibly embedded into each cover pixel quad, resulting in a
larger embedding rate up to 0.25 bit per pixel (bpp). Then, a sig-
nificant work based on difference expansion (DE) was reported in
[7] . By DE, in the best case, one bit can be embedded into a pixel
pair. Following [7] , many improvements have been proposed and
make RDH more efficient. Among these improvements, a good one
is [8] , in which the so called sorting technique is proposed to sort
pixel pairs according to their local variances before data embed-
∗
Corresponding author.
E-mail addresses: grxuan@tongji.edu.cn (G. Xuan), lixl@bjtu.edu.cn (X. Li),
shi@njit.edu (Y.-Q. Shi).
ding. By manipulating image histogram to embed data, another ef-
fective scheme called histogram shifting (HS) was reported [9,10] .
By HS, a good marked image quality is guaranteed with a suffi-
cient high embedding capacity. Later on, some improvements of HS
have been proposed [ 11 –14 ] so the performance of RDH is further
enhanced. Embedding data into prediction-errors is another sig-
nificant progress of RDH [15–17] . In this way, the prediction-error
histogram is firstly generated and RDH is implemented by modi-
fying the generated prediction-error histogram. Nowadays, embed-
ding data into prediction-errors is a major approach of RDH and
many recent works follow this way. For example, a very good em-
bedding result is obtained in [18] by using simultaneous sorting
and prediction-error histogram manipulation.
Some advanced RDH schemes have been reported in recent
years. In [19,20] , the theoretical analysis of [21] has been moved
further to achieve better performance in practical RDH. In [22] , the
method [20] is extended by considering a general distortion metric,
and several new embedding methods either for gray or for binary
image are presented. In [23] , a general construction for design-
ing HS-based RDH is proposed. In [24,25] , instead of the conven-
tional one-dimensional histogram, a two-dimensional histogram is
exploited for data embedding. In [26] , a dynamic histogram manip-
ulation method is proposed in which each prediction-error is adap-
tively modified according to its local context. In [27] , a multiple-
layer data embedding mechanism is proposed. In a recent work
[28] , a multiple histograms generation and modification strategy is
https://doi.org/10.1016/j.jisa.2018.01.006
2214-2126/© 2018 Elsevier Ltd. All rights reserved.
下载后可阅读完整内容，剩余9页未读，立即下载
2021-05-29 上传
2021-02-09 上传
2021-02-11 上传
2021-02-07 上传
2021-05-31 上传
2022-07-13 上传
2021-03-07 上传
weixin_38678300
- 粉丝: 4
- 资源: 1001
我的内容管理
展开
