Robust reversible watermarking scheme using Slantlet transform matrix
Graphical abstract
Introduction
Image watermarking techniques have been applied in many fields; some of these fields require recovering the original image after watermark extraction process because of the importance of this image. Medical imaging and military imaging are examples of the fields that give attention to the original image and the watermark at the same time. The type of image watermarking that is capable of recovering the original image is called reversible/lossless/invertible watermarking. Several reversible watermarking schemes have been proposed with different techniques (Alattar, 2004, An et al., 2009, Hwang et al., 2006, Kuo et al., 2007, Ni et al., 2006, Tian, 2002, Xuan et al., 2002, Xuan et al., 2005a, Xuan et al., 2005b, Xuan et al., 2006, Xuan et al., 2008).
Ni et al. (2006) proposed a histogram shifting reversible watermarking scheme. In this scheme, the embedding algorithm depends on finding the zero (or minimum) point and the peak (or maximum) point. The embedding process was done by shifting the grayscale values according to the points obtained and the watermark was embedded by modifying the maximum pixel values. The minimum and maximum points must be sent with the watermarked image as a side information. Hwang et al. (2006) proposed an improvement to the previous algorithm. In their work, a location map was implemented and embedded with the watermark; as a result, it is not necessary to send the side information to the receiver side. Kuo et al. (2007) improved the location map to increase the embedding capacity. In order to solve the problem of unstable performance of the previous methods, An et al. (2009) developed a histogram shifting based method. In this method, the authors proposed the use of a block statistical quantity (BSQ); because of the similarity of the histogram of these blocks, the diversity of grayscale histograms will be reduced. Tian (2002) proposed a reversible watermarking scheme based on difference expansion. The aim of his work is to obtain a reversible watermarking scheme with high capacity and high image quality. In order to increase the capacity of Tian's scheme, Alattar (2004) presented a hiding algorithm using an extended version of Tian's scheme by applying difference expansion (DE) to quads. The algorithm based on a reversible integer transform, which calculates the pairwise differences and the average between the elements of a vector extracted from the pixels of the image. Xuan et al. (2002) proposed a distortionless data hiding technique; in this technique, the embedding process was done in the middle and high frequency subbands by inserting a data bit into one (or more) middle bit plane(s) of the integer wavelet transform coefficients. To avoid the problem of overflow and underflow they used histogram modification. Xuan et al. (2005a) presented a lossless data hiding method in which the data embedding process depends on the comparison result of the high frequency integer wavelet coefficients with a specific threshold value. If the magnitude of the coefficient is less than the threshold, a binary bit can be embedded in the least significant bit-plane (LSB) of this coefficient. Histogram modification was applied as a pre-processing step to prevent overflow/underflow. In another method, Xuan et al. (2005b) proposed a watermarking scheme which depends on the compression and expansion of the wavelet coefficients. The histogram modification step was also used as a pre-processing to avoid underflow and overflow. Later, Xuan et al. (2006) proposed lossless watermarking technique based on histogram shifting in the transform domain. The histogram of the high frequency wavelet subband was shifted to give a space to insert the data. In another scheme, Xuan et al. (2008) proposed an improvement over the previous histogram shifting schemes which is aimed to get the best-watermarked image quality. The aim has been reached by choosing the best histogram threshold and applying the histogram adjustments only when it is necessary.
However, most of the proposed lossless data hiding schemes are fragile and the watermark is not robust for the actual applications. In these types of watermarking, the watermark cannot be recovered in case of unintentional attacks (e.g. noise addition, JPEG compression) and intentional attacks (e.g. active/passive attacks, collusion attacks). It is usually a regular way to use the lossy compression for saving the storage space required for images. Medical images are also compressed for saving the amount of bandwidth required for transmission of images in telemedicine environment. Thus, the image compression should be treated as incidental distortion. Here comes the need for a reversible watermarking method that can distinguish between unintentional and intentional attacks. The watermark that can withstand incidental distortion and can be destroyed by deliberate tampering is called semi-fragile (or robust) watermark.
Recently, a number of robust reversible watermarking methods have been proposed. De Vleeschouwer et al., 2001, De Vleeschouwer et al., 2003 proposed a robust lossless data hiding scheme based on using the grayscale histogram rotation. To avoid the problem of overflow/underflow, additions and subtractions with modulo-256 has been used. This method achieves good robustness against JPEG compression but the disadvantage of this technique is the salt-and-pepper noise in the stego images as pointed out by De Vleeschouwer himself. To prevent the salt-and-pepper noise effect, Zou et al., 2004, Zou et al., 2006 proposed a semi-fragile lossless watermarking scheme based on shifting the absolute mean values of the integer wavelet transform (IWT) coefficients. To avoid the overflow and underflow, the authors used spatial domain blocks classification and the error correction coding (ECC). Ni et al., 2004, Ni et al., 2008 designed a semi-fragile image authentication scheme which does not generate salt-and-pepper noise. The embedding process depends on applying the histogram distribution constrained in the spatial domain. Gao et al. (2009) developed an improved version of Ni's method to solve the problem of incomplete reversibility. An et al. (2010a) made a study of De Vleeschouwer et al. (2003), Zou et al. (2006) and Ni et al. (2008) methods, they found that Zou et al. (2006) and Ni et al. (2008) methods suffer from unstable reversibility and robustness. An et al. (2010b) proposed two robust reversible watermarking schemes based on using different strategies of reliable histogram rotation. The aim of those methods is to enhance the performance of De Vleeschouwer et al. (2003) method. Thereafter, An et al. (2012a) proposed a robust lossless data hiding using clustering and statistical quantity histogram. The aim of their work is to improve the visual quality and capacity. An et al. (2012b) proposed an improvement over the histogram rotation based embedding model of De Vleeschouwer et al. (2003) method. Their work was presented to solve the problem of salt-and-pepper noise and to improve the invisibility.
In this paper, we develop a new robust reversible watermarking method in the transform domain. This method is based on the use of the Slantlet transform (SLT) matrix to transform the image blocks and then the data bits are embedded into the SLT coefficients of a selected high frequency subband. The data embedding process is done by shifting the mean value of the chosen subband. To avoid the overflow and/or underflow, we used the histogram modification process based on Xuan et al. (2008) method; the reason of our choice to adopt the ideas of this histogram modification process is that to increase the visual quality of the watermarked image and to apply the pixel adjustment only when it is required. The purpose of the proposed method is to ensure the reversibility and to improve the robustness, capacity, and invisibility.
The rest of the paper is organized as follows. In Section 2, a brief description of the related works will be explained for better understanding of the proposed method. Section 3 contains the details of the proposed method. The experimental results and discussion are in Section 4. Section 5 contains the conclusion of the work done in this paper.
Section snippets
Related works
In this section, we introduce the Slantlet transform (SLT) (Selesnick, 1998, Selesnick, 1999) and the implementation of the SLT matrix. Then the semi-fragile lossless watermarking algorithm based on integer wavelet transform (IWT) method (Zou et al., 2006) will be explained.
The proposed method
In this section, we proposed a new robust reversible watermarking scheme. In a previous work, we found that the application of the SLT in image watermarking can give better image quality and more robustness as compared with the DWT based scheme. Inspired by this work and other applications of the SLT we suggested the use of the SLT instead of the integer wavelet transform (IWT) in the scheme proposed by Zou et al. (2006), but the IWT cannot be replaced directly by the SLT. We need to apply some
Experimental results and discussion
For a general comparison with the previous robust reversible watermarking schemes, first these schemes will be classified into two categories: the spatial domain schemes which are presented by De Vleeschouwer et al., 2001, De Vleeschouwer et al., 2003, Ni et al., 2004, Ni et al., 2008, Gao et al. (2009), An et al., 2010b, An et al., 2012a, An et al., 2012b and the transform domain schemes which are presented by Zou et al., 2004, Zou et al., 2006. The spatial domain schemes achieve higher
Conclusion
The main contribution of this paper is that a new robust reversible image-watermarking scheme is presented in order to obtain a completely reversible watermarking scheme with an improvement in the capacity, robustness, and invisibility. The main method includes: (1) dividing the host image into non-overlapping blocks and transforming these blocks using SLT matrix, (2) modifying the histogram of the spatial domain blocks in order to avoid the problem of overflow and/or underflow, and (3)
Rasha Thabit received her B.Sc. degree in Electronics and Communications Engineering from University of Baghdad, Iraq in 2006, and M.Sc. degree in Electrical Engineering from University of Baghdad, Iraq in 2008. She is currently a PhD. student in the School of Electrical & Electronic Engineering at Universiti Sains Malaysia (University of Science, Malaysia). Her research interest is in the area of digital image watermarking and signal processing.
References (39)
- et al.
Robust lossless data hiding using clustering and statistical quantity histogram
Neurocomputing
(2012) - et al.
Content-adaptive robust lossless data embedding
Neurocomputing
(2012) - et al.
Reversibility improved lossless data hiding
Signal Processing
(2009) - et al.
A Slantlet transform based intelligent system for magnetic resonance brain image classification
Elsevier, Biomedical Signal Processing and Control
(2006) - et al.
Slantlet transform-based OFDM scheme
Journal of Engineering
(2006) - et al.
Spoken word recognition using slantlet transform and dynamic time warping
Nahrain University, College of Engineering Journal (NUCEJ)
(2011) - et al.
ECG classification using slantlet transform and artificial neural network
Journal of Engineering
(2010) Reversible watermark using difference expansion of quads
- et al.
Reliable embedding for robust reversible watermarking
- et al.
Reversible watermarking based on statistical quantity histogram
Robust lossless data hiding: analysis and evaluation
Circular interpretation of histogram for reversible watermarking
Circular interpretation of bijective transformations in lossless watermarking for media asset management
IEEE Transactions on Multimedia
A reversible watermarking based on histogram shifting
Distortionless data hiding based on Slantlet transform
Reversible data hiding based on histogram
Watermarking in image using Slantlet transform
Iraqi Journal of Science
Image watermarking using Slantlet transform
Image compression using the Haar wavelet transform
Spelman Science and Math journal
Cited by (43)
Audio-lossless robust watermarking against desynchronization attacks
2022, Signal ProcessingCitation Excerpt :In case of no attacks, the distortion can be extracted for recovery of the original file. In the literature, researchers have designed various cover-lossless robust watermarking methods by using the kind of two stages of watermarking technique, e.g., for images [45–53] and audio files [54,55]. In these methods, few can resist geometric distortions or desynchronization attacks due to the fact that the robust watermark is sensitive to the spatial or time position.
A new approach of ECG steganography and prediction using deep learning
2021, Biomedical Signal Processing and ControlBirthmark based identification of software piracy using Haar wavelet
2019, Mathematics and Computers in SimulationCitation Excerpt :Many of the techniques are used for prevention of software piracy. Several versions of advanced techniques for theft detection is available such as software watermarking [28,43,48], finger prints [11,38], software cloning [6,39] plagiarism detection [1,44], and software birthmark [8,10,29,30,32,34,37,41]. Software watermark is used for the ownership of program, fingerprints is used for identification of intellectual property, software clone detection is used to detect piracy in code.
A novel blind tamper detection and localization scheme for multiple faces in digital images
2023, IET Image ProcessingRobust Reversible Image Watermarking based on Independent Embedding Domain and Pixel Value Ordering
2023, 8th International Conference on Recent Advances and Innovations in Engineering: Empowering Computing, Analytics, and Engineering Through Digital Innovation, ICRAIE 2023
Rasha Thabit received her B.Sc. degree in Electronics and Communications Engineering from University of Baghdad, Iraq in 2006, and M.Sc. degree in Electrical Engineering from University of Baghdad, Iraq in 2008. She is currently a PhD. student in the School of Electrical & Electronic Engineering at Universiti Sains Malaysia (University of Science, Malaysia). Her research interest is in the area of digital image watermarking and signal processing.
Bee Ee Khoo received the B.Tech degree in Quality Control and Instrumentation from Universiti Sains Malaysia in 1993, and PhD. degree in Electrical Engineering from University of Wales, Swansea in 1998. She is currently a senior lecturer with School of Electrical and Electronic Engineering, Universiti Sains Malaysia. Her current research interests include digital watermarking, computer vision and multimedia forensics.