基于小波变换的图像水印资源-CSDN文库

图像水印

需积分: 30 59 浏览量 2018-08-05 17:48:14 上传评论 1 收藏 6.05MB PDF 举报

资源推荐

资源详情

资源评论

Information Sciences 417 (2017) 381–400

Contents lists available at ScienceDirect

Information Sciences

journal homepage: www.elsevier.com/locate/ins

A new reliable optimized image watermarking scheme based

on the integer wavelet transform and singular value

decomposition for copyright protection

Nasrin M. Makbol

, Bee Ee Khoo

a , ∗

, Taha H. Rassem

, Khaled Loukhaoukha

School of Electrical and Electronic Engineering, Universiti Sains Malaysia, Malaysia

Faculty of Computer Systems and Software Engineering, Universiti Malaysia Pahang, Malaysia

Electrical Engineering and Computer Engineering Department, Laval University, Quebec, Canada

a r t i c l e i n f o

Article history:

Received 27 December 2016

Revised 22 April 2017

Accepted 14 July 2017

Available online 20 July 2017

Keywords:

Image watermarking

SVD

Integer wavelet transform

Multi-objective ant colony optimization

(MOACO)

False positive problem (FPP)

a b s t r a c t

Although image watermarking schemes based on singular value decomposition (SVD)

demonstrate high robustness and imperceptibility, they are exposed to the false positive

problem (FPP). This drawback mostly occurs when embedding steps depend on singular

values while singular vectors are used as secret keys. In this study, a new reliable SVD-

based image watermarking scheme that uses integer wavelet transform (IWT) is proposed

to overcome FPP and fulﬁl all watermarking requirements. Unlike in other schemes, the

S and V matrices of the watermark are used as secret keys, whereas the S singular vec-

tor of the watermark is embedded into the singular values of the host image. The ad-

ditional secret key is obtained from the watermarked image during the embedding pro-

cess to increase security and avoid FPP completely. To improve the robustness, as well as

achieve balance between robustness and imperceptibility, multi-objective ant colony opti-

mization (MOACO) is utilized to ﬁnd the optimal scaling factors, namely, multiple zooming

factors. Results of the robustness, imperceptibility, and reliability tests demonstrate that

the proposed IWT-SVD-MOACO scheme outperforms several previous schemes and avoids

FPP completely.

1. Introduction

Diverse computer and communication technologies have resulted in new opportunities to process and distribute multi-

media information. These technologies include powerful software and devices (e.g., scanners, printers, and digital cameras)

that enable users to create, manipulate, duplicate, and distribute images easily and economically. Moreover, the development

of multimedia networks and the availability of numerous content distribution applications on the Internet (e.g., peer-to-peer

ﬁle sharing and ﬁle transfer protocol) signiﬁcantly contribute to facilitating the exchange and distribution of multimedia con-

tent, even among unauthorised users [20] . Protecting the intellectual property rights (IPR) of content, verifying the origin

of content, and identifying authorised parties who initially distribute images to unauthorised parties are gaining attention.

Several techniques have been generated to address these issues. Watermarking technology is an approach to IPR protection.

∗

Corresponding author.

E-mail addresses: Nasrin_makbol@usm.my (N.M. Makbol), beekhoo@usm.my (B.E. Khoo), tahahussein@ump.edu.my (T.H. Rassem),

khaled.loukhaoukha.1@ulaval.ca (K. Loukhaoukha).

http://dx.doi.org/10.1016/j.ins.2017.07.026

382 N.M. Makbol et al. / Information Sciences 417 (2017) 381–400

Digital image watermarking protects content by embedding a signal; imperceptible information (i.e., copyright protection

information) into the host image without a noticeable degradation in visual quality. Consequently, a watermarked image is

developed and marked as public or sent to end users. Extracted or detected watermarks can be used for copyright protection

and content authentication [12,28] . Researchers interested in digital image watermarking are facing challenges in creating

new algorithms with suitable properties (requirements) to serve their intended applications. The essential requirements for

any watermarking technique are robustness, imperceptibility, capacity, and security. The performance of a given watermark-

ing scheme can be assessed based on these properties. Robustness is the capability of a scheme to maintain the validity of

including a watermark even after being subjected to geometric or non-geometric attacks. Imperceptibility is the property

to indicate that the watermarked data (image) should retain the quality of the original one as closely as possible. Capacity

refers to the number of bits that can be embedded into an image. A trade-off always exists among robustness, capacity,

and imperceptibility [8,12] . For example, enhancing watermark robustness will reduce its imperceptibility because of the

high watermark energy placed on a cover image [7] . Moreover, a high capacity will compromise imperceptibility because

numerous modiﬁcations in the cover image are required to embed the watermark. The fourth essential property, namely,

security, refers to the resistance of the scheme against hostile attacks. Invisible watermarks ensure that attackers cannot

access secured data to remove or alter them.

The current challenge is to achieve trade-off among the requirements of robustness, imperceptibility, and capacity. Most

Watermarking technologies prioritize robustness and imperceptibility, which are the major requirements that differentiate

watermarking from other data protection technologies [8,12] . The watermarking techniques are classiﬁed into spatial do-

main and transform domain techniques according to the embedding domain. Spatial domain techniques are the simplest;

however, they suffer from several drawbacks. They provide low embedding capacity and the watermark can be easily re-

moved using image processing operations. By contrast, transform domain techniques provide higher embedding capacity,

more robust, and can preserve the imperceptibility. They are gaining popularity and provide accurate aspects of the human

visual system model because of multi-resolution analysis [28] . Discrete wavelet transform (DWT) [1,18,23,34] , redundant

DWT (RDWT) [6,29] , integer wavelet transform(IWT) [30] , redistributed invariant discrete wavelet Transform (RIDWT) [4] ,

and lifting wavelet transform (LWT) [25] are examples of wavelet transforms.

Most image watermarking schemes improve their performance by combining two or more transforms; such schemes are

referred to as hybrid schemes. This idea concept is based on the assumption that combining two or more transforms can

make up for the defects of an individual transform, thereby resulting in an effective scheme [16,21,23,25] . The incentive for

developing hybrid schemes is to use the properties of the incorporated transforms and achieve the required goals of the

intended application. The success of hybrid schemes in achieving the desired goals depends on the appropriate selection of

the involved transforms. Several robust hybrids of digital-image watermarking schemes based on singular value decomposi-

tion (SVD) in the wavelet domain were developed a few years ago [6,23,29,30] . A matrix in SVD is decomposed into three

matrices: U, S and V

. These matrices have the same size as the original matrix as follows:

A = USV

(1)

where S represents the singular values, and U and V

are the left and right singular vectors, respectively. Various embed-

ding strategies of SVD-based watermarking schemes are based on the involved U, S , and V matrices. Most researchers prefer

to embed the watermark into S . Although the robustness and stability of such SVD-based watermarking schemes are good

when embedding is performed in S , these schemes face a high probability of encountering the false positive problem (FPP)

[23,29] . Recently, avoiding FPP has become one of the active research topics in the area of watermarking. FPP in SVD-based

watermarking schemes occurs because both the U and V matrices are adopted as secret keys. The suggested solution for

these kind of schemes is to propose an embedding technique that consider only one of the singular vectors as the secret

key. Furthermore, an optimization method is used to select multiple scaling factors (MSF) instead of a single scaling factor

(SSF) to maintain the trade-off between robustness and imperceptibility. Recently, artiﬁcial intelligence (AI) techniques have

played a major role in the watermarking ﬁeld to improve the performance of schemes by optimizing scaling factors. Several

optimization algorithms have been adopted and used in SVD-based watermarking schemes to achieve acceptable impercep-

tibility and robustness against most common attacks. Ant colony optimization (ACO), genetic algorithm (GA), differential

evolution (DE), artiﬁcial bee colony (ABC), and particle swarm optimization (PSO) are examples of optimization algorithms

for AI techniques [2–5,25–27,34,34,40] .

In this paper, a new optimized SVD-based image watermarking scheme in the IWT domain is proposed. We propose a

new embedding strategy where the singular vector ( U

) of the watermark is embedded into the singular values of the ﬁrst

level of the IWT host image ( S

). The novelty of this proposed scheme is using the singular vector ( V

) and singular values

( S

) as secret keys. Moreover, an extra secret key from the watermarked image is also generated in order to use it during

the extraction process to achieve more security and completely avoid the FPP. Multiple objective ACO (MOACO) algorithm is

used to select the optimal multiple zooming factors(MZF) to perform the proposed embedding strategy and maintain high

visual quality of the host image. This scheme is called as IWT-SVD-MOACO. The proposed scheme demonstrates high robust-

ness against different types of attacks as well as good imperceptibility. This scheme can solve the security issue attributed

to FPP, and thus, it can be used for copyright protection.

The rest of this paper is organized as follows. Section 2 offers an explanation of the FPP; its reasons, modes of attack, and

suggested solutions are presented. Section 3 brieﬂy describes the preliminaries used in the paper. In Section 4 , the proposed

IWT-SVD-MOACO scheme is explained. The experimental setup and results (i.e., imperceptibility and robustness tests, FPP

N.M. Makbol et al. / Information Sciences 417 (2017) 381–400 383

Table 1

Abbreviations that are used in the paper.

Abbreviations Meaning Abbreviations Meaning

SSF Single Scaling Factor MSF Multiple Scaling Factors

IPR Intellectual Property Rights AI Artiﬁcial Intelligence

FPP False Positive Problem SVD Singular Value Decomposition

IWT Integer Wavelet Transform DWT Discrete Wavelet Transform

RIDWT Redistributed Invariant Discrete Wavelet Transform

LWT Lifting Wavelet Transform RDWT Redundant DWT

MOACO Multi-Objective Ant Colony Optimization GA Genetic

Algorithm

PSO Particle Swarm Optimization ACO Ant Colony Optimization

ABC Artiﬁcial Bee Colony DE Differential Evolution

PSNR Peak Signal-to-Noise Ratio NC Normalized Correlation

reliability test and comparative analysis) of IWT-SVD-MOACO are presented in Section 5 . Finally, conclusions are provided in

Section 6 . The abbreviations that used in this paper are summarized in Table 1 .

2. False Positive Problem (FPP) in SVD-based image watermarking schemes

Due to the mathematical properties of SVD [11,22] , most SVD-based image watermarking schemes display high robust-

ness against image processing attacks and geometrical attacks while preserving good imperceptibility. Recently, SVD is pre-

ferred to be implemented with other transforms because it requires extensive computations when applied on its own. SVD-

based digital watermarking schemes embed a watermark by modifying either the singular values ( S ) or the singular vectors

( U and V ). Each embedding method has advantages and disadvantages. Most researchers prefer to embed into S given the

stability and properties of S . Despite the stability and robustness of SVD-based image watermarking schemes that embed the

watermark by altering S , these schemes fail to resolve the issue of rightful ownership because they are vulnerable to FPP. In

this problem, a counterfeit watermark is detected from a content which a different watermark [16,23,24] . Two embedding

methods lead to such problem. The ﬁrst type of embedding method that is vulnerable to FPP is identiﬁed as follows:

S + αW = U

W Host

(2)

where S is the singular value matrix of the host image; W is the watermark; and U

WHost

, S

WHost

, and V

W Host

are the matrices

obtained after applying SVD to the result of the previous embedding operation. ( S

WHost

) represents the singular values of the

watermarked image (i.e., the image in which the watermark has been embedded), whereas ( U

WHost

, V

W Host

) are the singular

vectors. Examples of schemes that adopts such embedding method are [23,29] . The second method is identiﬁed as follows:

Water mar ked

= S + αS

water mar k

(3)

In this embedding strategy, the watermark is subjected to SVD, and then U

watermark

, S

watermark

, and V

water mar k

are obtained.

S and, S

watermark

are the singular values of the host and the watermark, respectively. The embedding process is achieved by

altering S by adding S

watermark

to it after multiplying it with a scaling factor ( α). S

Watermarked

represents the watermarked

data. Ganic and Eskicioglu [16] , and Rastegar et al. [33] proposed schemes that followed the embedding method described

in Eq. (3) .

The observed ﬂaws of the aforementioned embedding methods that are addressed using Eqs. (2) and (3) result from using

( U and V ) as side information or secret keys for the extraction process. The U and V matrices include major information on

the structure of an image. Thus, FPP occurs because an attacker can claim rightful ownership by using these secret keys in

different manners (attacks). Three potential attacks are applied to any watermarking scheme using Eq. (3) or Eq. (2) as the

embedding method. These attacks have been discussed and addressed in [35,41,42] . To explain how these attacks operate,

they are tested on the embedding method indicated by Eq. (2) . Moreover, similar attacks may occur using the embedding

method described in Eq. (3) . The attacks are as follows according to Guo and Prasetyo [17] :

1. Attack 1

This type of attack ( Fig. 1 ) is interpreted as an ambiguity on the part of the owner. Suppose the owner has O as the

host image and two different watermarks W

and W

. The owner applies the watermark embedding method twice. In

the ﬁrst, W

is embedded into O , which yields O

as the watermarked image and ( U

and V

) as the secret keys. In

the second, W

is embedded into O , which yields O

as the watermarked image and U

and V

as the secret keys. FPP

occurs during the extraction process when the owner extracts a watermark W

2 ∗

from the O

using the incorrect secret

keys ( U

and V

). W

2 ∗

is visually similar to W

. This result is considered to be a corrupted watermark because W

is the

correct embedded watermark in O

. A similar problem may occur when a watermark W

1 ∗

is extracted from O

using

the incorrect secret keys ( U

and V

. Thus, the owner extracts the corrupted watermark indicated as W

1 ∗

, which appears

similar to W

. However, the correct embedded watermark is W

2. Attack 2

剩余19页未读，继续阅读

评论收藏

内容反馈

jmunet

粉丝: 1
资源: 16

基于小波变换的图像水印

基于小波变换的图像水印系统的研究与开发

小波变换(Wavelet)实现数字水印

matlab利用小波变换进行数字图像水印添加与提取

基于小波变换的数字水印技术

使用matlab实现的小波变换彩色图像水印嵌入和提取程序

基于深度学习的小波转化的数字水印去除算法

matlab图像添加水印

基于小波变换的数字水印嵌入与提取_matlab代码

小波分析 水印 经典实例

基于小波变换的图像数字水印MATLAB代码

基于小波变换的图像拼接

基于小波变换的图像分割

基于小波变换的数字水印方法,基于小波变换的数字水印技术,Python源码.zip

基于小波变换的数字水印算法

基于小波变换的数字图像水印技术

基于小波变换的小波水印代码.zip

基于小波变换的数字水印嵌入与提取算法

基于小波变换的数字水印源代码

基于小波变换的数字水印MATLAB代码

基于小波变换的数字水印方法_python_数字水印python_数字水印python_水印_小波变换

最新资源

小波分析水印经典实例