THÔNG TIN TÀI LIỆU
Signals and Communication Technology Rohit M. Thanki Advanced Techniques for Audio Watermarking Signals and Communication Technology Series Editors: Emre Celebi Department of Computer Science University of Central Arkansas Conway, AR, USA Jingdong Chen Northwestern Polytechnical University Xi’an, China E. S. Gopi Department of Electronics and Communication Engineering National Institute of Technology Tiruchirappalli, Tamil Nadu, India Amy Neustein Linguistic Technology Systems Fort Lee, NJ, USA H. Vincent Poor Department of Electrical Engineering Princeton University Princeton, NJ, USA This series is devoted to fundamentals and applications of modern methods of signal processing and cutting-edge communication technologies The main topics are information and signal theory, acoustical signal processing, image processing and multimedia systems, mobile and wireless communications, and computer and communication networks Volumes in the series address researchers in academia and industrial R&D departments The series is application-oriented The level of presentation of each individual volume, however, depends on the subject and can range from practical to scientific “Signals and Communication Technology” is indexed by Scopus More information about this series at http://www.springer.com/series/4748 Rohit M. Thanki Advanced Techniques for Audio Watermarking Rohit M. Thanki C. U Shah University Wadhwan City, Gujarat, India ISSN 1860-4862 ISSN 1860-4870 (electronic) Signals and Communication Technology ISBN 978-3-030-24185-8 ISBN 978-3-030-24186-5 (eBook) https://doi.org/10.1007/978-3-030-24186-5 © Springer Nature Switzerland AG 2020 This work is subject to copyright All rights are reserved by the Publisher, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed The use of general descriptive names, registered names, trademarks, service marks, etc in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use The publisher, the authors, and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication Neither the publisher nor the authors or the editors give a warranty, express or implied, with respect to the material contained herein or for any errors or omissions that may have been made The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations This Springer imprint is published by the registered company Springer Nature Switzerland AG The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland Preface Digital watermarking is an important technique because digital multimedia data such as digital images, digital videos, and audio signals, that is shared over the Internet This book intends to provide basic information and overviews of various techniques for security of audio signals It is written for students, researchers, and professionals who work in security domain and want to improve the knowledge in this course Digital watermarking technique can be used in copyright protection and authentication of multimedia content Thus, content cannot be modified or altered by someone The need of security for audio signals exists everywhere due to its rapid distribution over the Internet or social media This book aims to provide basic technical information on audio watermarking to students, researchers, and professionals The concept of audio watermarking in this book is given with the help of figures and graphs, so that the readers can easily get concept of theories and techniques Several watermarking algorithms are presented in this book with figures and its fundamental principles for audio signals The combination of data encryption and watermarking technique provides a new concept for security of audio signal The bio-inspired based watermarking technique solves the manual selection of gain factor and provides an optimized embedding approach The book also covers the application of audio watermarking technique for security of speech signal in the biometric system Overview of the Book In Chap 1, basic information and properties of audio watermarking are briefly discussed The rest of the book covers various audio watermarking techniques in Chaps 2, 3, 4, 5, and In Chap 2, the basic mathematical preliminaries are discussed including various signal transforms, such as discrete cosine transform, discrete wavelet transform, singular value decomposition, curvelet transform, contourlet transform, and ridgelet transform, and various encryption methods, such as chaotic map and compressive sensing (CS) based, random noise sequence v vi Preface generation, audio watermarking attacks, and evaluation parameters for audio watermarking The fundamentals of audio watermarking are covered in Chap It is divided into three classes: spatial domain watermarking, transform domain watermarking, and hybrid domain watermarking The advanced audio watermarking techniques are discussed in Chap The techniques based on curvelet transform, ridgelet transform, and contourlet transform for audio signal are introduced and discussed In Chap 5, the technique of combination of audio watermarking and encryption is introduced The combination of various encryption methods with audio watermarking technique is described, and its advantages and experimental results are explained Chapter shows how optimization algorithm can be applied in audio watermarking The concept and basic operations of particle swarm optimization and genetic algorithm and the fitness function are also discussed Then, optimization- based audio watermarking is introduced The summary of the book is discussed in Chap Following the overview, the experiment results are also presented Features of the Book • • • • New state-of-the-art algorithms for audio watermarking Several practical results of algorithms Extensive discussion on advanced audio watermarking algorithms Inclusion of optimization-based audio watermarking Acknowledgments My task has been easier and the final version of the book has considerably been better because of the help I have received Acknowledging that help is itself a pleasure I would extend many thanks to all the persons who helped me achieve the final version of this book I am indebted to numerous colleagues for their valuable suggestions during the entire period of the manuscript preparation I would also like to thank the publishers at Springer, in particular Mary James, senior publishing editor, for their helpful guidance and encouragement during the creation of this book Wadhwan City, Gujarat, India Rohit M. Thanki Contents 1 Introduction�������������������������������������������������������������������������������������������������� 1 2 Mathematical Preliminaries������������������������������������������������������������������������ 7 3 Fundamental of Audio Watermarking ������������������������������������������������������ 25 4 Blind Audio Watermarking ������������������������������������������������������������������������ 47 5 Audio Watermarking with Encryption������������������������������������������������������ 59 6 Optimization-Based Audio Watermarking������������������������������������������������ 83 7 Summary of Book���������������������������������������������������������������������������������������� 97 Index�������������������������������������������������������������������������������������������������������������������� 99 vii List of Figures Fig 1.1 Basic structure of digital watermarking system���������������������������������� 2 Fig 1.2 Classification of watermarking ���������������������������������������������������������� 3 Fig 2.1 Test audio signals: (a) pop, (b) classical, (c) jazz, (d) loopy music ���������������������������������������������������������������������������������� 8 Fig 2.2 DCT coefficients of audio signal�������������������������������������������������������� 9 Fig 2.3 DWT coefficients of audio signal: (a) approximation wavelet coefficients, (b) details wavelet coefficients�������������������������� 11 Fig 2.4 SVD coefficients of audio signal: (a) coefficients of U matrix, (b) coefficients of S matrix, (c) coefficients of V matrix�������������������� 12 Fig 2.5 High-frequency curvelet coefficients of audio signal ������������������������ 13 Fig 2.6 Finite ridgelet transform for signal ���������������������������������������������������� 14 Fig 2.7 Ridgelet coefficients of audio signal�������������������������������������������������� 14 Fig 2.8 (a) Original audio signal, (b) low-frequency contourlet coefficients of signal, (c–g) high-frequency contourlet coefficients of signal �������������������������������������������������������������������������� 15 Fig 2.9 Process for CS-based encryption and decryption ������������������������������ 18 Fig 2.10 Generation of encrypted image using CS-based encryption method: (a) original image, (b) encrypted image, (c) decrypted image���������������������������������������������������������������������������� 18 Fig 2.11 (a) Watermarked audio signal, (b) after resampling attack, (c) after additive noise attack, (d) after filtering attack, (e) after cropping attack���������������������������������������������������������������������� 21 Fig 3.1 Embedding process of LSB substitution for audio watermarking in spatial domain���������������������������������������������������������� 26 Fig 3.2 Extraction process of LSB substitution for audio watermarking in spatial domain���������������������������������������������������������� 27 Fig 3.3 Embedding process of additive audio watermarking in spatial domain �������������������������������������������������������������������������������� 28 ix x List of Figures Fig 3.4 Extraction process of additive audio watermarking in spatial domain �������������������������������������������������������������������������������� 28 Fig 3.5 Simulation results of additive audio watermarking in spatial domain �������������������������������������������������������������������������������� 29 Fig 3.6 DCT coefficients with size of 8 × 8���������������������������������������������������� 31 Fig 3.7 Simulation results of DCT-based substitution audio watermarking�������������������������������������������������������������������������������������� 32 Fig 3.8 Simulation results of DCT-based multiplicative audio watermarking�������������������������������������������������������������������������������������� 35 Fig 3.9 Simulation results of DWT-based multiplicative audio watermarking�������������������������������������������������������������������������������������� 38 Fig 3.10 Simulation results of SVD-based additive audio watermarking�������������������������������������������������������������������������������������� 40 Fig 3.11 Simulation results of DWT + SVD-based hybrid audio watermarking�������������������������������������������������������������������������������������� 44 Fig 4.1 Concept of audio watermarking with noise sequences ���������������������� 48 Fig 4.2 Simulation results of correlation-based audio watermarking in spatial domain �������������������������������������������������������������������������������� 49 Fig 4.3 Block diagram for SWT-based audio watermarking�������������������������� 51 Fig 4.4 Simulation results of SWT-based audio watermarking���������������������� 53 Fig 4.5 Simulation results of FDCuT- and DCT-based audio watermarking�������������������������������������������������������������������������������������� 56 Fig 4.6 Recovered watermark images: (a) resampling attack, (b) additive noise attack, (c) filtering attack, (d) cropping attack������������������������������������������������������������������������������ 57 Fig 5.1 Audio watermarking with encryption: (a) first approach and (b) second approach �������������������������������������������������������������������� 61 Fig 5.2 Block diagram of DCT + FDCuT + SVD-based audio watermarking: (a) embedding process and (b) extraction process ������������������������������������������������������������������ 62 Fig 5.3 Simulation results of DCT + FDCuT + SVD-based audio watermarking���������������������������������������������������������������������������� 65 Fig 5.4 Extracted watermark images for DCT + FDCuT + SVD-based audio watermarking: (a) resampling attack, (b) additive noise attack, (c) filtering attack, and (d) cropping attack�������������������� 66 Fig 5.5 Simulation results of DCT + FDCuT + SVD-based audio watermarking�������������������������������������������������������������������������������������� 69 Fig 5.6 Encryption process in FDCuT- and SWT-based audio watermarking: (a) original watermark image, (b) scrambled watermark image, (c) extracted watermark image, and (d) extracted scrambled watermark image ���������������������������������� 69 Fig 5.7 Extracted watermark images for FDCuT- and SWT-based audio watermarking: (a) resampling attack, (b) additive noise attack, (c) filtering attack, and (d) cropping attack ���������������������������� 70 86 6 Optimization-Based Audio Watermarking 6.3 Working of Optimization and Bioinspired Algorithms In this section, the working of optimization and bioinspired algorithms such as genetic algorithm (GA), particle swarm optimization (PSO), and simulated annealing (SA) is discussed 6.3.1 Genetic Algorithm (GA) Genetic algorithm (GA) was developed by Holland et al [14] It is a searching technique based on the concept of Darwin’s theory of natural evolution This is a direct search method that depends on natural section machines and efficiently works for a large number of users and finds an optimum solution [16, 17] The GA has the following advantages compared to traditional optimization methods: This algorithm works on code value of parameters instead of the actual value of parameters The traditional methods work on single point search, while GA works on multiple point searches This approach improves the find global optimal value and provides more robustness This algorithm does not use any auxiliary value of problem parameters Hence, this algorithm is only applicable to continuous or discrete optimization problems It uses probabilistic transfer function, while traditional optimization method uses deterministic transfer function The GA uses a population of classes that give optimal solutions Each class is evaluated based on some fitness function measurement to optimal solution form the optimization problem This algorithm is robust, flexible, and efficient on various type spaces when optimal solution searches for the optimization problem It is not a simple random search optimization method, but it utilizes knowledge of the previous stage and generates a new optimal solution There are six steps in this algorithm: (1) problem identification, (2) initialization of class, (3) evaluation of fitness function (4) constraint handling (5) generation of the new class, and (6) stopping criteria The basic flow chart of GA is shown in Fig. 6.1 The first important step of applying GA to the optimization problem is the encoding method because it sets window limitation for information that has used in this system In GA, the information represents in multiple chromosomes The chromosome is a string of variables and called a gene This variable is binary numbers and real numbers, and its length determined the problem specification The two parameters such as class and process for initialized the class are used in GA for initialization of the class GA generates multiple class points with the predefined size of the class This gives the GA to search for multiple different probabilities of the problem space and results in the global optimal solution The two methods such as random 6.3 Working of Optimization and Bioinspired Algorithms 87 Fig 6.1 Flow chart of genetic algorithm (GA) initialization and heuristic initialization are used in GA. After initialization of class, GA uses the survival principle of nature to the search process and uses the fitness function value as input information to determine the space for the problem GA is naturally used for solving maximization problems In this problem, the fitness function is first derived from the optimization problem and used in a successive process of GA. GA is most suitable for unconstrained optimization problems, but most of the problems are constrained in nature Thus, the first constrained problem is converted into an unconstrained problem In this approach, one additional penalty function is added to the optimization problem This function can be added by two approaches such as based on the number of constraints violated and based on some distance from the flexible region This function has some characteristics such as it should be progressive and factor of this function is summarized values of all loss done due to constraints violated The generation of a new class is done using different operators such as selection, crossover, and mutation The selection method is a selected class from multiple classes according to their fitness function The fitness function of each class is calculated with respect to a given optimization problem Once the selection process is over, the crossover is applied Crossover is a recombination operator that combines sub-information of two main chromosomes to produce new information that contains some information of both main chromosomes The above two operators are generating a large number of data strings that create two problems such as GA 88 6 Optimization-Based Audio Watermarking searches the entire space of optimization problem due to less diversity in the initial data strings and GA may have sub-optimum strings due to the wrong choice of initial class These problems are overcome by the mutation operator in GA. This is used to inject new genetic data into the genetic classes In this process, the parent string can either replace the whole class or replace less fit value in a string During operation of GA, fitness function value increases gradually, and at particular condition, the increment in fitness function is not possible, and this value represents the optimal or near optimal solution value At this stage, the operation of GA is terminated The example of crossover operation and mutation operation is shown in below Fig. 6.2 6.3.2 Particle Swarm Optimization (PSO) PSO algorithm is used to solve multidimensional optimization problems and proposed by Kennedy et al [15] This algorithm is inspired by the social behavior of flocking of bird or fish schooling This algorithm is very similar to GA. In this algorithm, initially, a class of particles is generated randomly, and then the optimal value is calculated using the iterative search method Here, a velocity vector, as well as position vector, is calculated for each particle Based on these vectors, the fitness function is calculated, and based on this value, the best solution in the swarm is found for each particle Best particle form these all local give global optimal particles The basic equations for the PSO algorithm are given as [18, 19] z = α ∗ z + C1 ∗ rand ( x − p ) + C2 ∗ rand ( y − p ) p = p+z (6.3) (6.4) where β is a user-defined inertial weight parameter, C1 and C2 are the acceleration weights and control the previous values of the particle velocities on its current one, and rand is a random number whose value lies in the interval of [0 1] The new velocity of the particle is calculated using Eq. 6.3 based on the velocity of the previous particle, the collaborative effect of all particles, and the distance between the current position and best historical position of the particle The new position of particles is updated using Eq. 6.4 The flow chart of the PSO algorithm is given in Before Crossover Operation Parent string 1: 0010 1010 Parent string 2: 1101 1101 Before Mutation Operation Parent string 1: 0010 1010 Parent string 2: 1101 1101 Fig 6.2 Example of crossover and mutation operation After Crossover Operation Parent string 1: 0010 1101 Parent string 2: 1101 1010 After Mutation Operation Parent string 1: 0110 1000 Parent string 2: 0101 1001 6.3 Working of Optimization and Bioinspired Algorithms 89 Fig. 6.3 According to the literature [17, 18], the inertial weight parameter can be calculated as β= imax − i imax (6.5) where β is user-defined inertial weight parameter and imax is maximum iteration value The value of inertial weight lies in the interval of [0 0.99] Fig 6.3 Flow chart of particle swarm optimization (PSO) algorithm 90 6 Optimization-Based Audio Watermarking 6.3.3 Simulated Annealing (SA) Simulated annealing (SA) is a local search optimization algorithm [20–24] This algorithm is based on annealing phenomena of science The annealing is a thermal process to finding low energy states of an atom in a heated environment The process contains two steps: increase the temperature of the heat environment to a maximum value at which the atom melts, and decrease the temperature of heat environment carefully until the particles arrange themselves in the ground state condition of the atom This state has a minimum energy state of the atom The value of this state can be obtained only if the maximum temperature is high enough and the cooling is done slowly The connection between the annealing process and optimal minimization was established by Pincus et al [22] The annealing process as an optimization technique is proposed by Kirkpatrick et al [23] for the combinational optimization problem The SA-based optimization process can be performed using the Metropolis algorithm [22, 24] which is based on the Monte Carlo method This algorithm is generating an optimal solution to combinational optimization problems by assuming an analogy between the input function and physical many particle systems with the following assumptions: (a) solution of the problem is equivalent to states of a physical system and (b) the value of a solution is equivalent to the “energy” of a state For implementation of SA-based algorithm for the solution of the optimization process, below functions are required: A successor function that returns a “close” neighboring solution given the actual optimal value This function will work as a distributive function for the particles of the system A target function to optimize that depends on the current state of the system This function will work as the energy of the system The main advantage of SA is able to avoid becoming trapped at local minima [25] The algorithm uses a random search method which accepts both changes such as increment and decrement in input function f The probability of optimal value for this algorithm is given as δf p = exp − T (6.6) where δf is the changes in input function and T is a control parameter The implementation of the SA algorithm is very early The basic flow chart of the SA algorithm is given in Fig. 6.4 [25] The input parameters of SA algorithm is possible solutions value, generation of random changes in solutions, a mean value of evaluating the problem functions, initial temperature, and method or rules for decreasing it in the search process Only one image watermarking technique was available in the literature [26] This technique is used for the calculation of the optimal scaling factor 91 6.4 Optimization-Based Audio Watermarking Fig 6.4 Flow chart of simulated annealing (SA) algorithm 6.4 Optimization-Based Audio Watermarking In this section, the audio watermarking technique using GA, PSO, and SA with its experimental results is discussed Here, watermark embedding and extraction are done using DCT-based substitution audio watermarking The steps for watermark embedding and extraction for this technique are described in Chap Here, DCT coefficients of the host audio signal are modified by the watermark image and optimized scaling factor [26–28] The fitness function for each optimization algorithms can be calculated using obtained NC values The maximum fitness values are selected and treated as the best optimal solution The fitness function is given by the following equation: NC = {corr ( AS,WAS ) + corr ( w, w ∗)} / Fitness = − NC (6.7) 92 6 Optimization-Based Audio Watermarking In this above equation, Fitness is a fitness function; AS and WAS indicate host audio signal and watermarked audio signal, respectively; w and w* indicate original watermark image and extracted watermark image, respectively The experimental results show that this fitness function works well for optimized audio watermarking technique The simulation results of GA-based optimized audio watermarking are shown in Fig. 6.5 where (a) shows the original pop audio signal, (b) shows watermarked pop audio signal, (c) shows the original watermark image, and (d) shows the extracted watermark image Here, the size of the original audio signal has samples of 65,536 with 1 second duration, and size of the watermark image is 32 × 32 pixels Figure 6.6 shows the quality of the extracted watermark image using GA-based optimized audio watermarking against audio watermarking attacks From Fig. 6.6, Fig 6.5 Simulation results of GA-based optimized audio watermarking 6.4 Optimization-Based Audio Watermarking 93 it is indicated that this technique provides good robustness against Additive Noise Attack and has less robustness against Filtering Attack Figure 6.7 shows the quality of the extracted watermark image using PSO-based optimized audio watermarking against audio watermarking attacks From Fig. 6.7, it is indicated that this technique provides less robustness against Resampling Attack and Filtering Attack Figure 6.8 shows the quality of the extracted watermark image using SA-based optimized audio watermarking against audio watermarking attacks From Fig. 6.8, it is indicated that this approach provides good robustness against Additive Noise Attack The performance measurement of optimized audio watermarking is summarized in Table 6.2 From Table 6.2, it is indicated that the SA-based optimized technique provides good perceptual transparency compared to other two optimization algorithms Fig 6.6 Extracted watermark images for GA-based optimized audio watermarking: (a) Resampling Attack, (b) Additive Noise Attack, (c) Filtering Attack, (d) Cropping Attack Fig 6.7 Extracted watermark images for PSO-based optimized audio watermarking: (a) Resampling Attack, (b) Additive Noise Attack, (c) Filtering Attack, (d) Cropping Attack Fig 6.8 Extracted watermark images for SA-based optimized audio watermarking: (a) Resampling Attack, (b) Additive Noise Attack, (c) Filtering Attack, (d) Cropping Attack 6 Optimization-Based Audio Watermarking 94 Table 6.2 Performance measurement of optimized audio watermarking using GA, PSO, and SA Attack SNR (dB) No Attack Resampling Attack Additive Noise Attack Filtering Attack Cropping Attack 11.2949 6.3332 3.7635 2.7685 1.4179 No Attack Resampling Attack Additive Noise Attack Filtering Attack Cropping Attack 11.2949 –1.4469 4.8385 2.7848 1.4179 No Attack Resampling Attack Additive Noise Attack Filtering Attack Cropping Attack 11.2949 7.6962 4.8501 2.7848 1.4179 NC (a) Using GA 1.0000 0.7035 1.0000 0.6590 0.9852 (b) Using PSO 1.0000 0.8019 0.8315 0.5067 0.9852 (c) Using SA 1.0000 0.5081 0.8208 0.5067 0.9852 BER 0.0000 0.2430 0.0000 0.3540 0.1720 0.0000 0.1470 0.1730 0.4970 0.1720 0.0000 0.4980 0.1740 0.4970 0.1720 References Dhar P, Shimamura T (2015) Advances in audio watermarking based on singular value decomposition SpringerBriefs in Electrical and Computer Engineering, Springer, Germany Xiang Y, Hua G, Yan B (2017) Digital audio watermarking: fundamentals, techniques and challenges Springer, Singapore Lin Y, Abdulla WH (2015) Audio watermark, vol 146 Springer, Heidelberg Cvejic N (ed) (2007) Digital audio watermarking techniques and technologies: applications and benchmarks: applications and benchmarks IGI Global, USA Euler L (2014) Optimization techniques: an overview In: Multidimensional particle swarm optimization for machine learning and pattern recognition, pp 13–44, Springer, Germany Optimization methods, http://www.cse.iitm.ac.in/~vplab/courses/optimization/Optim_methods.pdf Last access: Jan 2018 Venter G (2010) Review of optimization techniques In: Encyclopedia of aerospace engineering, Wiley, USA Fletcher R, Reeves CM (1964) Function minimization by conjugate gradients Comput J 7(2):149–154 Head JD, Zerner MC (1985) A Broyden—Fletcher—Goldfarb—Shanno optimization procedure for molecular geometries Chem Phys Lett 122(3):264–270 10 Fiacco AV, McCormick GP (1968) Nonlinear programming: sequential unconstrained minimization techniques Wiley, USA 11 Chen TY (1993) Calculation of the move limits for the sequential linear programming method Int J Numer Methods Eng 36(15):2661–2679 12 Nocedal J, Wright SJ (2006) Sequential quadratic programming Springer, New York, pp 529–562 References 95 13 Borra S, Thanki R, Dey N (2018) Digital image watermarking: theoretical and computational advances CRC Press, USA 14 Holland JH (1992) Adaptation in natural and artificial systems: an introductory analysis with applications to biology, control, and artificial intelligence MIT Press, USA 15 Kennedy J (2010) Particle swarm optimization In: Encyclopedia of machine learning, pp 760– 766, Springer, US 16 Goldberg DE (1989) Genetic algorithms in search, optimization, and machine learning Addison-Wesley, Reading 17 Pelikan M, Goldberg DE, Lobo FG (2002) A survey of optimization by building and using probabilistic models Comput Optim Appl 21(1):5–20 18 Binitha S, Sathya SS (2012) A survey of bio inspired optimization algorithms Int J Soft Comput Eng 2(2):137–151 19 Waleed J, Jun HD, Abbas T, Hameed S, Hatem H (2014) A survey of digital image watermarking optimization based on nature inspired algorithms NIAs Int J Secur Appl 8(6):315–334 20 Aarts EHL, Korst JHM, Arbib MA (2003) Simulated annealing and Boltzmann machines In: Handbook of brain theory and neural networks, 2nd edn, pp 1039–1044, MIT Press, USA 21 Otten RH, van Ginneken LP (2012) The annealing algorithm, vol 72 Springer Science & Business Media, Springer, US 22 Pincus M (1970) Letter to the editor—a Monte Carlo method for the approximate solution of certain types of constrained optimization problems Oper Res 18(6):1225–1228 23 Kirkpatrick S, Gelatt CD, Vecchi MP (1983) Optimization by simulated annealing Science 220(4598):671–680 24 Metropolis N, Rosenbluth AW, Rosenbluth MN, Teller AH, Teller E (1953) Equation of state calculations by fast computing machines J Chem Phys 21(6):1087–1092 25 Simulated Annealing Website: https://www.phy.ornl.gov/csep/mo/node29.html Last access: Mar 2018 26 Lin GS, Chang YT, Lie WN (2010) A framework of enhancing image steganography with picture quality optimization and anti-steganalysis based on simulated annealing algorithm IEEE Trans Multimedia 12(5):345–357 27 Cox IJ, Miller ML, Bloom JA, Honsinger C (2002) Digital watermarking, vol 53 Morgan Kaufmann, San Francisco 28 Cox IJ, Kilian J, Leighton T, Shamoon T (1996) Secure spread spectrum watermarking for images, audio and video In: Proceedings of 3rd IEEE international conference on image processing, vol IEEE, pp 243–246, Lausanne, Switzerland Chapter Summary of Book Audio watermarking is a process of inserting a secret watermark image within the original audio signal to show copyright and authenticity of the owner It has been utilized effectively to provide solutions for copyright protection, privacy protection, and content authentication The audio watermarking techniques presented in this book are summarized as follows: • In Chap 3, the fundamental watermarking techniques in the spatial domain, transform domain, and hybrid domain for audio signals are discussed Here, various techniques such as LSB substitution-based technique, DCT substitution- based technique, DCT multiplicative technique, DWT-based multiplicative technique, and SVD-based additive technique are presented with its experimental results The hybrid domain audio watermarking is also discussed in this chapter LSB substitution-based technique and multiplicative audio watermarking techniques not provide good robustness against various audio watermarking attacks, while SVD-based additive technique and hybrid domain audio watermarking technique provide good robustness against various audio watermarking attacks • In Chap 4, the blind watermarking techniques for audio signals are presented Here, advanced blind audio watermarking techniques using SWT and FDCuT are discussed with its experimental results These techniques provide good robustness against various audio watermarking attacks • An audio watermarking technique with encryption has been presented in Chap Here, two encryption methods such as Arnold scrambling and CS based in audio watermarking are presented The various audio watermarking with Arnold scrambling based on the combination of FDCuT, DCT, SVD, and combination of FDCuT and SVD are presented The audio watermarking in the encryption domain is also discussed in this chapter The FDCuT-based audio watermarking does not provide robustness against various audio watermarking attacks Therefore, this technique is used owner authentication of the audio signal against piracy © Springer Nature Switzerland AG 2020 R M Thanki, Advanced Techniques for Audio Watermarking, Signals and Communication Technology, https://doi.org/10.1007/978-3-030-24186-5_7 97 98 7 Summary of Book • The use of optimization in audio watermarking is discussed in Chap Here, various optimization algorithms such as genetic algorithm (GA), particle swarm optimization (PSO), and simulated annealing (SA) with its working are presented The optimized audio watermarking using DCT and optimization has been presented with its experimental results in this chapter The optimized audio watermarking provides good robustness against additive noise attack • The presented audio watermarking techniques can be effectively used for various applications such as copyright protection, privacy protection, and ownership authentication There are several directions for future research in this presented area introduced in this book In the future work, more optimization-based audio watermarking will be proposed to improve the selection of optimized scaling factor In addition, the performance of optimized audio watermarking further improves against watermarking attacks The robustness of audio watermarking techniques against advanced audio watermarking attacks such as channel fading, jitter, and packet drop needs to be checked Index A Additive audio watermarking, 27–29, 37 Additive noise attack, 93, 94 Additive white Gaussian noise (AWGN), 20 Advanced audio watermarking techniques, Advanced Encryption Standard (AES), Arnold scrambling, 14, 17, 59, 60 DCT + FDCuT + SVD embedding process, 61, 62 extracted watermark images, 66 extraction process, 62, 63 non-blind audio watermarking technique, 61 performance measurement of, 64, 66 simulation results, 64, 65 FDCuT and SWT, 64–71 Audio matrix (AM), 65, 76 Audio signal, 1, 2, 7, 28 Audio watermarking technique, 7, 10, 11, 13, 20, 41, 47 B Bioinspired algorithms (BIA) overview, 84–85 working of GA, 86–88 PSO, 88–89 SA, 90–91 Bit error rate (BER), 22, 84, 94 Blind audio watermarking technique, 64, 75, 97 Blocking effect, C Cartesian polar transform (CPT), 59 Compressive sensing (CS)-based encryption, 14, 60, 68–80 Compressive sensing (CS) theory, 17 Continuous wavelet transform (CWT), 10 Copyright authentication, 74 Correlation-based audio watermarking noise sequences, 49 performance measurement, 50 simulation results, 49 Cropping attack, 20, 93, 94 CS-based encryption process, 17 D Data encryption methods Arnold scrambling, 17 CS theory, 17 Destination-based watermarking, Digital audio signals, Digital watermarking, Discrete cosine transform (DCT), 5, 30, 52, 59–66, 69 audio watermarking, multiplicative audio watermarking extraction, 34 performance measurement, 36 simulation results, 34 substitution audio watermarking advantage, 31 coefficients, 31 performance measurement, 33 simulation results, 32 © Springer Nature Switzerland AG 2020 R M Thanki, Advanced Techniques for Audio Watermarking, Signals and Communication Technology, https://doi.org/10.1007/978-3-030-24186-5 99 100 Discrete time curvelet transform (DTCuT), 13 Discrete wavelet transform (DWT), 5, 10, 30 DWT-based multiplicative audio watermarking audio watermarking type, 37 performance measurement, 37 simulation results, 37 steps, 36 watermark extraction, 36 DWT + SVD-based hybrid audio watermarking, 42, 43 E Evolutionary-based optimization algorithms, 85 Evaluation parameters FNE, 23 FPE, 23 watermark images, 22 F False-negative error (FNE), 23, 68, 71 False-positive error (FPE), 23, 68, 71 Fast discrete curvelet transform (FDCuT), 52, 60–75 Fast Fourier transform (FFT), 59 FDCuT-and DCT-based audio watermarking, 56 Filter Attack, 20, 93, 94 Finite ridgelet transform (FRT), 13 First approach, 60, 61 Fragile, 74 G Gaussian noise (WGN), 49 Genetic algorithm (GA), 83, 85–88, 98 H Hiding grayscale watermark image, 60 High-frequency curvelet coefficients (HCu), 63, 64, 71 Host audio signal (HAS), 59, 60, 64, 70, 73, 75, 76 I Inverse stationary wavelet transform (ISWT), 51 Invisible watermarking, Index L Least significant bit (LSB), 26, 59 M MATLAB toolbox, 13 Multiplicative audio watermarking non-blind approach, 34 scaling factor, 33 Multiplicative watermarking, 33 N Noise sequences additive, 47 concept, 48 correlation-based audio watermarking, 49 technique, 47 Non-blind audio watermarking technique, 61 Non-subsampled contourlet transform (NSCT), 14 Normalized correlation (NC), 22, 84 O Optimization audio watermarking fitness function, 92 GA-based, 92–94 performance measurement, 94 PSO-based, 93, 94 SA-based, 93, 94 scaling factor, 91 simulation results, 92 technique, 91 bioinspired algorithms (see Bioinspired algorithms) bioinspired-based, 83 need, 83–84 scaling factor, 83 P Particle swarm optimization (PSO), 83, 85, 86, 88–89, 98 Payload capacity, Pop audio signal, 13 Pseudorandom noise (PN), 19, 49, 66 Pseudorandom noise generator, 19 Index R Resampling attack, 93, 94 Robustness, Robust watermarking algorithms, S Scaling factor, 83, 90, 91 Second approach, 60, 61 Secret watermark image, 97 Signal to noise ratio (SNR), 20, 84, 94 Simulated annealing (SA), 83, 85, 86, 90–91, 98 Singular value decomposition (SVD), 4, 10, 30, 37 Source-based watermarking, Spatial domain audio watermarking disadvantage, 25 extraction side, 26 LSB, 26 type, 25 Spatial domain watermarking, Stationary wavelet transform (SWT), 50 block diagram, 51 performance measurement, 52 simulation results, 52, 53 watermark extraction, 51 wavelet coefficients, 50 Structural similarity index measure (SSIM), 84 Substitution-based audio watermarking, 33 Substitution-based watermarking, 31 SVD-based additive audio watermarking, 39–41 101 T Techniques, audio watermarking Arnold scrambling (see Arnold scrambling) CS-based encryption domain, 75–80 FDCuT, 68–75 data encryption, 60–61 hybrid watermarking techniques, 59 Traditional audio watermarking techniques, Transform domain audio watermarking audio signal, 31 DCT, 30 DWT, 36 extraction side, 30 Transform domain watermarking, 4, 33 V Visible watermarking, 4, W Watermark audio matrix (WAM), 48, 67, 77 Watermarked audio signal (WAS), 67, 73, 77 Watermarking application, classification, copyright protection, Watermarking technique, 14 White Gaussian noise (WGN), 19, 52, 54 ... India ISSN 186 0-4 862 ISSN 186 0-4 870 (electronic) Signals and Communication Technology ISBN 97 8-3 -0 3 0-2 418 5-8 ISBN 97 8-3 -0 3 0-2 418 6-5 (eBook) https://doi.org/10.1007/97 8-3 -0 3 0-2 418 6-5 © Springer... Springer Nature Switzerland AG 2020 R M Thanki, Advanced Techniques for Audio Watermarking, Signals and Communication Technology, https://doi.org/10.1007/97 8-3 -0 3 0-2 418 6-5 _1 1 Introduction Fig 1.1 ... Springer Nature Switzerland AG 2020 R M Thanki, Advanced Techniques for Audio Watermarking, Signals and Communication Technology, https://doi.org/10.1007/97 8-3 -0 3 0-2 418 6-5 _2 (2.2) 2 Mathematical
Ngày đăng: 08/05/2020, 06:41
Xem thêm: