Biometrics inspired watermarking based on a fractional dual tree complex wavelet transform

https://doi.org/10.1016/j.future.2012.05.021Get rights and content

Abstract

In this paper, a novel biometrics inspired watermarking technique is proposed. For this purpose, a newly proposed mathematical transform, namely the fractional dual tree complex wavelet transform (FrDT-CWT), and singular value decomposition (SVD) are used. The core idea is to use biometrically generated keys in the embedding process of a gray-scale watermark. Therefore, this paper first proposes a method for generating keys from biometrics efficiently. The host image is first randomized by Hessenberg decomposition and a chaotic map, followed by embedding in the FrDT-CWT domain by modifying the singular values of the randomized image. Further, in order to prevent the ambiguity problem of SVD-based techniques, a verification step is introduced to verify the watermarked image. Finally, a reliable extraction process is proposed to extract the watermark from the possibly attacked watermarked image after verification. The security, attack, and comparative analysis confirms the high security, efficiency, and robustness of the proposed watermarking technique.

Highlights

► A novel biometrics inspired watermarking technique is proposed. ► The watermark is embedded in a fractional dual tree complex wavelet transform domain using SVD. ► A method is also proposed to generate keys from biometrics efficiently. ► To prevent the ambiguity problem of SVD-based techniques, a verification step is casted. ► The security, attack, and comparative analysis confirms the high security, efficiency, and robustness of the proposed technique.

Introduction

Both the success and the substantial proliferation of web technologies have created an environment in which some crucial issues for digital media such as illegal copying, distribution, editing, and authentication have become very easy. The phenomenon has led to an increasing need for developing some standard solutions to prevent these issues. One of the technical solutions is to provide law enforcement and copyright protection for digital media which can be achieved practically by digital watermarking. Digital watermarking refers to a technique that imperceptibly inserts an authorized mark information (watermark) into the digital media. This hidden information can be retrieved by the contrary process for a variety of purposes [1]. In recent years, several watermarking algorithms have been proposed in the literature. These algorithms can be broadly classified in two categories, according to the embedding domain: the spatial domain and the transform domain. Spatial domain approaches [2] are the simplest, and the earliest algorithms were based on the modification of pixel intensities. These algorithms are generally fragile to numerous attacks. On the other hand, transform domain approaches insert the watermark into the transform coefficients; examples include the Fourier transform [3], cosine transform [4], wavelet transform [5], [6], [7], [8], fractional Fourier transform [9], [10], [11], and dual-tree complex wavelet transform [12].

Recently, a new transform, a singular value decomposition (SVD)-based [13], [14] watermarking technique, and its variants have been proposed. These approaches work on the simple concept of finding the SVD of a cover image or the SVD of each block of the cover image and then modifying the singular values to embed the watermark. Further, some researchers have presented hybrid watermarking schemes in which they have combined SVD with the other existing transforms [15], [16], [17], [18], [19], [20], [21]. The main reason behind the hybridization is the fact that SVD-based scheme withstands a variety of attacks but it is not resistant to geometric attacks such as rotation, cropping, etc. Hence, to improve the performance, hybridization is needed. It was however first argued by Zhang and Li [22] that SVD-based schemes fail under ambiguity attacks, i.e. by taking recourse to the reference matrices of the watermark, the same can be extracted from a possibly distorted watermarked image, and this fact is again proved in [23]. The fact that SVD subspace can preserve major information of an image leads to the above-mentioned drawback. Thereafter, other authors have also discussed the same drawback, and have given its solution to some extent [17], [24], [25], [26].

The main motive of this work is to develop and implement a new concept in an SVD-based hybrid watermarking scheme which cannot fail under ambiguity attacks. The core idea of the possible solution is to introduce a key concept in the SVD-based scheme such that, even if someone has knowledge of the full embedding process, then without these keys he/she can never extract the watermark. In the proposed work, these keys are generated by the biometrics of the owner/user of the digital media. Recently, biometrics has emerged as a promising technology for personal identification and authentication [27], [28]. Biometrics is gaining increasing interest of research as well as corporate communities due to its highly secure and trustworthy characteristics. Generally, biometrics refers to methods that can be used to uniquely recognize individuals based upon one or more intrinsic physical or behavioral characteristics. In information technology, in particular, biometrics is used as a tool for efficient and reliable identity management and access control [29]. Therefore, the development of a digital media security system is proposed in this work which will use the biometrics as an actuating factor to strengthen the security. The key concept is introduced by the recently proposed fractional transform, namely the fractional dual tree complex wavelet transform (FrDT-CWT) [30]. The FrDT-CWT is a realization of the DT-CWT in the fractional Fourier domain. The FrFT has a unique property of describing the information of the spatial and frequency domain due to the rotation of the time–frequency plane over an arbitrary angle. In contrast, the DT-CWT has a multiresolution property. A combination of these two domain results in the FrDT-CWT, which exhibits the multiresolution property, describing the spatial as well as the frequency domain information. The transform orders of the FrDT-CWT act as the key in the proposed scheme and therefore are generated by the biometrics of the owner/user. Therefore, an efficient way to generate keys from biometrics images is also suggested which is based on the speeded-up robust features (SURF) technique. The SURF technique transforms an object into a large collection of local feature vectors, each of which is invariant to translation, scaling, and rotation, affine, or three-dimensional (3D) projection, and partially invariant to illumination changes.

  • This paper proposes the use of a recently introduced transform, namely the fractional dual tree complex wavelet transform, for the aim of secure digital image watermarking.

  • This paper proposes an efficient way to generate keys and transform orders from the biometrics of the user/owner.

  • A new invertible randomization process for gray-scale images is proposed exploiting the characteristics of a nonlinear chaotic map and Hessenberg decomposition.

  • Further, this technique is an attempt to rectify the drawback of SVD in image watermarking by introducing a verification step based on Tsallis entropy.

The remainder of the paper is organized as follows. In Section 2, an introduction to a nonlinear chaotic map, Hessenberg decomposition, singular value decomposition, and SURF is given. The definition of the fractional dual tree complex wavelet transform is given in Section 3, followed by the proposed biometrics inspired watermarking technique in Section 4. The experimental results and security analysis are briefly described in Section 5. Finally, the concluding remarks are given in Section 6.

Section snippets

Mathematical preliminaries

In this section, the main terminologies are given which are used in the proposed biometrics inspired watermarking technique to achieve the desired goal. These terminologies are as follows.

Fractional dual tree complex wavelet transform

The fractional dual tree complex wavelet transform (FrDT-CWT) is a realization of the dual tree complex wavelet transform in the fractional Fourier domain [30]. The fractional Fourier transform has a unique property of describing the information of the spatial and frequency domain due to the rotation of the time–frequency plane over an arbitrary angle. In contrast, the dual tree complex wavelet transform has a multiresolution property. A combination of these two domains result in the FrDT-CWT,

Proposed biometrics inspired watermarking technique

In this section, some of the motivating factors in the design of our approach to the biometrics inspired watermarking technique are discussed. Without loss of generality, assume that I and W represent the original host and watermark images of size M×N and M1×N1, respectively, such that the watermark image is smaller than the host image by a factor 2q1 and 2q2 along both directions, where q1 and q2 are any positive integers greater than or equal to 1. The core idea is to capture the appropriate

Experimental setup

The performance of the proposed biometrics inspired watermarking framework is demonstrated using a MATLAB platform. A number of experiments are performed on different gray-scale images of size 256 × 256, namely Boat, Lena, Mandril, Lady, Balloon, and Montage. Also, six different gray-scale images of size 64 × 64, namely Logo1, Logo2, Logo3, Logo4, Logo5, and Logo6, respectively, are used as the watermark images. Watermarks Logo1, Logo2, Logo3, Logo4, Logo5, and Logo6 are embedded into Boat,

Conclusions

In this paper, a novel watermarking technique based on a fractional dual tree complex wavelet transform and singular value decomposition is proposed in which the key concept is introduced and implemented by the biometrics of the owner/user to ensure better security. Therefore, a new and efficient method is suggested to generate keys from the biometrics of the owner/user. The main benefit of using biometrics-based keys is that no one can obtain the keys without the information of the biometrics

Acknowledgments

The work was supported by the Canada Research Chair program, the Natural Sciences and Engineering Research Council of Canada (NSERC) Discovery Grant.

Further, the authors thank the anonymous referees and the Editor for the valuable suggestions and many constructive comments that resulted in improvements and better readability of this paper.

Gaurav Bhatnagar has been a member of the Computer Vision and Sensing Systems Laboratory in the Department of Electrical and Computer Engineering at the University of Windsor, ON, Canada, since 2009. He received his Ph.D. and M.Sc. degrees in Applied Mathematics from the Indian Institute of Technology Roorkee, India, in 2010 and 2005, respectively. He has co-authored more than 30 journal articles and conference proceedings, and has contributed to two books in his area of interest. His research

References (46)

  • G.C. Langelaar et al.

    Watermarking digital image and video data

    IEEE Signal Processing Magazine

    (2000)
  • I.J. Cox et al.

    Secure spread spectrum watermarking for multimedia

    IEEE Transactions on Image Processing

    (1997)
  • M. Barni et al.

    Improved wavelet based watermarking through pixel wise masking

    IEEE Transactions on Image Processing

    (2001)
  • Y. Wang et al.

    A wavelet based watermarking algorithm for ownership verification of digital images

    IEEE Transactions on Image Processing

    (2002)
  • D. Kundur et al.

    Towards robust logo watermarking using multiresolution image fusion

    IEEE Transactions on Multimedia

    (2004)
  • Z. Feng, M. Xiaomin, Y. Shouyi, Multiple-chirp typed blind watermarking algorithm based on fractional Fourier...
  • C. Delong, Dual digital watermarking algorithm for image based on fractional Fourier transform, in: Proc. of...
  • Joong-Jae Lee et al.

    A new incremental watermarking based on dual-tree complex wavelet transform

    Journal of Supercomputing

    (2005)
  • R. Liu et al.

    An SVD-based watermarking scheme for protecting rightful ownership

    IEEE Transactions on Multimedia

    (2002)
  • D.V.S. Chandra, Digital image watermarking using singular value decomposition, in: Proc. IEEE Midwest Symposium on...
  • E. Ganic et al.

    Robust embedding of visual watermarks using discrete wavelet transform and singular value decomposition

    Journal of Electronic Imaging

    (2005)
  • A. Sverldov, S. Dexter, A.M. Eskicioglu, Robust DCT–SVD domain image watermarking for copyright protection: embedding...
  • Gaurav Bhatnagar et al.

    Robust reference—watermarking scheme using wavelet packet transform and bidiagonal–singular value decomposition

    International Journal of Image and Graphics

    (2009)
  • Cited by (0)

    Gaurav Bhatnagar has been a member of the Computer Vision and Sensing Systems Laboratory in the Department of Electrical and Computer Engineering at the University of Windsor, ON, Canada, since 2009. He received his Ph.D. and M.Sc. degrees in Applied Mathematics from the Indian Institute of Technology Roorkee, India, in 2010 and 2005, respectively. He has co-authored more than 30 journal articles and conference proceedings, and has contributed to two books in his area of interest. His research interests include digital watermarking, encryption techniques, biometrics, image analysis, wavelet analysis, and fractional transform theory.

    Q.M. Jonathan Wu received his Ph.D. degree in electrical engineering from the University of Wales, Wales, UK, in 1990. From 1995, he was with the National Research Council of Canada, Ottawa, ON, Canada, for ten years, where he became a Senior Research Officer and Group Leader. He is currently a Full Professor in the Department of Electrical and Computer Engineering, University of Windsor, Windsor, ON. He holds the Tier 1 Canada Research Chair in Automotive Sensors and Sensing Systems. He has published more than 200 peer-reviewed papers in the areas of computer vision, image processing, intelligent systems, robotics, micro- sensors and actuators, and integrated micro-systems. His current research interests include 3D computer vision, active video object tracking and extraction, interactive multimedia, sensor analysis and fusion, and visual sensor networks. Dr. Wu is an Associate Editor for the IEEE Transaction on Systems, Man, and Cybernetics (part A). He is on the editorial board of the International Journal of Robotics and Automation. He is a member of the IEEE Technical Committee on Robotics and Intelligent Sensing. He has served on the Technical Program Committees and International Advisory Committees for many prestigious international conferences.

    View full text