Robust watermarking based on DWT and nonnegative matrix factorization
Wei Lu
a,
*
, Wei Sun
a
, Hongtao Lu
b
a
School of Information Science and Technology, Guangdong Key Laboratory of Information Security Technology, Sun Yat-sen University, Guangzhou 510275, China
b
Department of Computer Science and Engineering, Shanghai Jiao Tong University, Shanghai 200240, China
article info
Article history:
Received 16 September 2007
Received in revised form 1 July 2008
Accepted 16 September 2008
Available online 5 November 2008
Keywords:
Watermarking
Nonnegative matrix factorization
DWT
abstract
This paper presents a novel blind robust digital image watermarking scheme using non-
negative matrix factorization (NMF) in DWT domain. Firstly, the original image is trans-
formed into some subband coefficients using discrete wavelet transformation (DWT),
and then a Gaussian pseudo-random watermark sequence is embedded in the factorized
decomposition coefficients using NMF. Because of the multiresolution decomposition for
DWT and physically meaningful factorization for NMF, the proposed scheme can achieve
good robustness, which is also demonstrated in the following experiments.
Ó 2008 Elsevier Ltd. All rights reserved.
1. Introduction
In the past two decades, digital watermarking technology has been devoted significantly, and applied in digital right pro-
tection and authentications widely. On one hand, many different approaches are introduced into digital watermarking, such
as signal processing, pattern recognition, communication theory, etc., which have improved the performance of watermark-
ing. On the other hand, most of the current watermarking schemes are still not applied in practice, since the performance of
these watermarking algorithms is still far away from practical applications, especially for robustness, which is the most
important design target.
These years, a new signal decomposition method is proposed as nonnegative matrix factorization (NMF) [1,2], which
decomposes a nonnegative matrix into two physically meaningful nonnegative matrices. This method has been applied in
many signal analysis domain [3], such as frontal face verification [4], blind signal separation [5], image classification [6].
To our current knowledge, some applications for NMF in digital watermarking have been developed [7–11].In[11], a robust
image watermarking scheme is developed using DWT and NMF, which apply NMF to the blocks of each wavelet decompo-
sition subband, and then is followed by eigendecomposition distortion. This scheme achieves a better visual imperceptibility
and excellent robustness against intentional and geometric attacks. However, this scheme is not blind, that is to say the
watermark detection scheme needs the original cover image that is not provided in most cases. In this paper, we propose
a blind digital image watermarking scheme using NMF in DWT domain based on Spread Spectrum technology. It is blind
and does not need original image in the watermark detection process. Through embedding a pseudo-random sequence in
the NMF coefficient matrix in DWT domain, the watermarking scheme can resist many signal attacks and distortions, and
thus achieve strong robustness.
The rest parts of this paper are organized as follows. In Section 2, we introduce the NMF process briefly. In Section 3, the
proposed watermarking scheme is described in detail. Then, the experimental results are shown in Section 4. Finally, con-
clusions are given in Section 5.
0045-7906/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.compeleceng.2008.09.004
* Corresponding author.
E-mail addresses: luwei3@mail.sysu.edu.cn (W. Lu), sunwei@mail.sysu.edu.cn (W. Sun), lu-ht@cs.sjtu.edu.cn (H. Lu).
Computers and Electrical Engineering 35 (2009) 183–188
Contents lists available at ScienceDirect
Computers and Electrical Engineering
journal homepage: www.elsevier.com/locate/compeleceng