#### DMCA

## Data hiding in image and video: Part I—Fundamental issues and solutions (2003)

Venue: | IEEE Trans. Image Processing |

Citations: | 21 - 8 self |

### Citations

1072 | T.: Secure spread spectrum watermarking for multimedia
- Cox, Lilian, et al.
- 1997
(Show Context)
Citation Context ... Earlier works regarding the embedding capacity focused on spread spectrum additive watermarking, by which a noise-like watermark is added to a host image and is later detected via a correlator [19], =-=[20]-=-. This embedding can be modeled as communication over a channel with additive white gaussian noise (AWGN) [21], [22]. Other researchers studied the bounds of embedding capacity under blind detection [... |

1024 |
Writing on dirty paper
- Costa
- 1983
(Show Context)
Citation Context ...tion [23]–[25]. Zero-error capacity has been studied for a watermark-based authentication system under magnitude-bounded noise [17], using the principles originally proposed by Shannon [26], [27]. In =-=[28]-=-, Costa showed that the channel capacity under two additive Gaussian noises with one known to the sender equals to the capacity in the absence of the known noise. This result has been incorporated in ... |

626 | A.: Techniques for data hiding - Bender, Gruhl, et al. - 1996 |

490 | Quantization index modulation: A class of provably good methods for digital watermarking and information embedding
- Chen, Wornell
- 2001
(Show Context)
Citation Context ...e Gaussian noises with one known to the sender equals to the capacity in the absence of the known noise. This result has been incorporated in information theoretical formulations of data hiding [29], =-=[30]-=-. The gap between the theoretical embedding capacity in data hiding and what is achievable in practice can be bridged by investigation of such issues as basic embedding mechanisms for embedding one bi... |

356 |
The zero-error capacity of a noisy channel
- Shannon
- 1956
(Show Context)
Citation Context ...der blind detection [23]–[25]. Zero-error capacity has been studied for a watermark-based authentication system under magnitude-bounded noise [17], using the principles originally proposed by Shannon =-=[26]-=-, [27]. In [28], Costa showed that the channel capacity under two additive Gaussian noises with one known to the sender equals to the capacity in the absence of the known noise. This result has been i... |

340 | Introduction to watermarking techniques - Kutter, Hartung - 2000 |

278 |
An Introduction to Probability Theory and Its Applications. 3rd ed
- Feller
- 1968
(Show Context)
Citation Context ...nts will not contribute much to this investigation. The values from (8) and (9) are (12) The derivation is given in the Appendix . The expected histogram { } is a hypergeometric distribution function =-=[47]-=- and can be approximated well by a binomial distribution with mean . An excelor by Poisson and normal distributions with mean lent approximation of is [18] (9) (10) (11) These quantities depend only o... |

261 | Information-theoretic analysis of information hiding
- Moulin, O'Sullivan
- 1999
(Show Context)
Citation Context ...dditive Gaussian noises with one known to the sender equals to the capacity in the absence of the known noise. This result has been incorporated in information theoretical formulations of data hiding =-=[29]-=-, [30]. The gap between the theoretical embedding capacity in data hiding and what is achievable in practice can be bridged by investigation of such issues as basic embedding mechanisms for embedding ... |

148 | Information hiding—a survey - Petitcolas, Anderson, et al. - 1999 |

143 | Towards robust and hidden image copyright labeling
- Koch, Zhao
- 1995
(Show Context)
Citation Context ... partition the signal space. For example, for a pair of host samples or coefficients and , we may generate the marked version and so that a “1” is embedded by forcing and a “0” is embedded by forcing =-=[36]-=-. We can also use enforcing signs to embed one bit [37], [38]. Many schemes that use noncoherent detection3 belong to this Type-II category. It is the deterministically enforced relationship on that r... |

128 |
An Invisible Watermarking Technique for Image Verification
- Yeung, Mintzer
- 1997
(Show Context)
Citation Context ... host components. The odd-even embedding can be viewed as a special case of the table-lookup embedding, which can provide an additional level of security by using a random lookup table as the mapping =-=[34]-=-, [35]. There are many other ways to partition the signal space. For example, for a pair of host samples or coefficients and , we may generate the marked version and so that a “1” is embedded by forci... |

104 | Anti-collusion fingerprinting for multimedia
- Trappe, Wu, et al.
- 2003
(Show Context)
Citation Context ...gonal and biorthogonal modulations is exponential with respect to , and thus inefficient except for small 8 . 8 A divide-and-conquer detection algorithm for orthogonal modulation recently proposed in =-=[42]-=- can reduce the computation complexity from O(2 ) to O(B) at an expense of detection accuracy.690 IEEE TRANSACTIONS ON IMAGE PROCESSING, VOL. 12, NO. 6, JUNE 2003 There is considerable freedom in sel... |

102 | B.O.Comiskey, “Modulation and Information Hiding in Images
- Smith
- 1996
(Show Context)
Citation Context ...oise-like watermark is added to a host image and is later detected via a correlator [19], [20]. This embedding can be modeled as communication over a channel with additive white gaussian noise (AWGN) =-=[21]-=-, [22]. Other researchers studied the bounds of embedding capacity under blind detection [23]–[25]. Zero-error capacity has been studied for a watermark-based authentication system under magnitude-bou... |

98 | and B.Liu. Watermarking for Image Authentication
- Wu
- 1998
(Show Context)
Citation Context ...components. The odd-even embedding can be viewed as a special case of the table-lookup embedding, which can provide an additional level of security by using a random lookup table as the mapping [34], =-=[35]-=-. There are many other ways to partition the signal space. For example, for a pair of host samples or coefficients and , we may generate the marked version and so that a “1” is embedded by forcing and... |

97 |
Random allocations
- Kolchin, Sevastyanov, et al.
- 1978
(Show Context)
Citation Context ...1) Equalizing Embedding Capacity Via Shuffling: The fraction of segments having embeddable coefficients is . Computing the exact distribution of is quite involved. Instead, we adopt a moment approach =-=[46]-=- to study the mean and variance of each normalized bin of the histogram. As shall be seen, studying higher moments is not necessary. For each bin of the histogram where ,wehave (8) Fig. 10. Histogram ... |

76 |
Image adaptive watermarking using visual models
- Podilchuk, Zeng
- 1998
(Show Context)
Citation Context ...ce in detection. Although it is possible to detect directly from [32], the knowledge of will enhance detection performance by eliminating the interference. Additive spread spectrum watermarking [20], =-=[33]-=- is an example of Type-I. In Type-II embedding, the signal space is partitioned into subsets, each of which is mapped by a function to the set of values taken by the secondary data, as illustrated in ... |

54 | Fast robust template matching for affine resistant image watermarks
- Pereira, Pun
- 1999
(Show Context)
Citation Context ...ction V-C. Shuffling increases the sensitivity to such intentional attacks as geometric distortion. This sensitivity can be alleviated through registration with respect to a known reference watermark =-=[50]-=-, [51]. Besides applying shuffling to the entire set of samples or coefficients, we can shuffle on block basis by permuting all samples/ coefficients in the same block as a whole [52]. We can also app... |

52 | The first 50 years of electronic watermarking
- Cox, Miller
- 2002
(Show Context)
Citation Context ...tication, access control, and annotation. Data hiding is also found to be useful to send side information in multimedia communication for achieving additional functionalities or enhancing performance =-=[3]-=-. In this paper, we address both fundamental and design issues regarding data hiding in image and video. The paper is organized into two parts. Part I addresses several fundamental issues and proposes... |

52 | Multimedia Data Hiding
- Wu, Liu
- 2002
(Show Context)
Citation Context ...gy era cannot be fully realized without the guarantee on the security and protection of multimedia data. Digital watermarking and data hiding1 are schemes to embed secondary data in digital media [2]–=-=[18]-=- for a variety of applications, including ownership protection, authentication, access control, and annotation. Data hiding is also found to be useful to send side information in multimedia communicat... |

52 | A statistical watermark detection technique without using original images for resolving rightful ownerships of digital images
- Zeng, Liu
- 1999
(Show Context)
Citation Context ... embed one bit , the difference between marked signal and the original host signal is a function of , i.e., . can be a major noise source in detection. Although it is possible to detect directly from =-=[32]-=-, the knowledge of will enhance detection performance by eliminating the interference. Additive spread spectrum watermarking [20], [33] is an example of Type-I. In Type-II embedding, the signal space ... |

51 | Effective and ineffective digital watermarks - Mintzer, Braudaway, et al. - 1997 |

51 | Robust data hiding for images
- Swanson, Zhu, et al.
- 1996
(Show Context)
Citation Context ...r bit is reduced as increases, causing a decrease in detection reliability. For Type-II, the embedding of multiple bits can be done by enforcing relations along several mutually orthogonal directions =-=[43]-=-, [44]. B. Comparison of Modulation/Multiplexing Techniques Applicability: The applicability of a particular technique depends on the type of multimedia sources and the embedding mechanism being used.... |

42 |
The Art of Computer Programming, 3rd ed
- Knuth
- 1998
(Show Context)
Citation Context ...rmation. The detector only needs to know the segment size and the shuffle table. A shuffling table can be generated efficiently from a key with linear complexity proportional to the number of entries =-=[48]-=-. 2) Practical Considerations: While our analysis shows that the probability for getting a bad shuffle that cannot equalize the uneven capacity is extremely small, it is still possible for a given ima... |

41 |
Data hiding in digital binary image
- Wu, Tang, et al.
- 2000
(Show Context)
Citation Context ...nontrivial. For example, it is difficult to find in a binary image many overlapped but orthogonal directions to produce features that are manipulable within the just-noticeable-difference (JND) range =-=[53]-=-, [54]; to obtain such directions for audio also requires a large window of samples, which could lead to significant processing delay. TDM versus CDM: TDM and orthogonal CDM are equivalent in terms of... |

39 | Design and Analysis of Digital Watermarking, Information Embedding, and Data Hiding Systems - Chen - 2000 |

33 | Capacity issues in digital image watermarking
- Servetto, Podilchuk, et al.
- 1998
(Show Context)
Citation Context ...ike watermark is added to a host image and is later detected via a correlator [19], [20]. This embedding can be modeled as communication over a channel with additive white gaussian noise (AWGN) [21], =-=[22]-=-. Other researchers studied the bounds of embedding capacity under blind detection [23]–[25]. Zero-error capacity has been studied for a watermark-based authentication system under magnitude-bounded n... |

33 |
Hidden signatures in image
- Hsu, Wu
- 1996
(Show Context)
Citation Context ... host samples or coefficients and , we may generate the marked version and so that a “1” is embedded by forcing and a “0” is embedded by forcing [36]. We can also use enforcing signs to embed one bit =-=[37]-=-, [38]. Many schemes that use noncoherent detection3 belong to this Type-II category. It is the deterministically enforced relationship on that removes the need of using the host signal . For convenie... |

24 | Data hiding in multimedia: Theory and applications - Ramkumar - 1998 |

21 | Information theoretic bounds for data hiding in compressed images
- Ramkumar, Akansu
- 1998
(Show Context)
Citation Context ...]. This embedding can be modeled as communication over a channel with additive white gaussian noise (AWGN) [21], [22]. Other researchers studied the bounds of embedding capacity under blind detection =-=[23]-=-–[25]. Zero-error capacity has been studied for a watermark-based authentication system under magnitude-bounded noise [17], using the principles originally proposed by Shannon [26], [27]. In [28], Cos... |

17 | Multiresolution digital watermarking: algorithms and implications for multimedia signals, doctoral thesis - Kundur - 1999 |

17 | A Bayesian approach to affine transformation resistant image and video
- Csurka, Deguillaume, et al.
- 1999
(Show Context)
Citation Context ...V-C. Shuffling increases the sensitivity to such intentional attacks as geometric distortion. This sensitivity can be alleviated through registration with respect to a known reference watermark [50], =-=[51]-=-. Besides applying shuffling to the entire set of samples or coefficients, we can shuffle on block basis by permuting all samples/ coefficients in the same block as a whole [52]. We can also apply dif... |

16 | Multi-level data hiding for digital image and video
- Wu, Yu, et al.
- 1999
(Show Context)
Citation Context ...host media is basic to every data hiding system. Almost all embedding approaches belong Fig. 2. Layered structure of data hiding. to one of two general types, which was proposed independently in [29]–=-=[31]-=-. In Type-I embedding, the secondary data, possibly encoded, modulated, and/or scaled, is added to the host signal, as illustrated in Fig. 3(a). The addition can be performed in a specific domain or o... |

16 | Iteratively decodable codes for watermarking applications
- Kesal, Mihcak, et al.
- 2000
(Show Context)
Citation Context ...stronger than the watermark. Motivated by Costa’s information theoretical result [28], distortion compensation has been proposed to be incorporated into quantization-based enforcement embedding [30], =-=[39]-=-, [40], where the enforcement is combined linearly with the host signal to form a watermarked signal. The optimal scaling factor is a function of WNR and will increase the number of bits that can be e... |

15 | Multimedia Security and Copyright Protection , Ph - Qiao - 1998 |

12 |
Capacity of full frame DCT image watermarks
- Barni, Bartolini, et al.
(Show Context)
Citation Context ...is embedding can be modeled as communication over a channel with additive white gaussian noise (AWGN) [21], [22]. Other researchers studied the bounds of embedding capacity under blind detection [23]–=-=[25]-=-. Zero-error capacity has been studied for a watermark-based authentication system under magnitude-bounded noise [17], using the principles originally proposed by Shannon [26], [27]. In [28], Costa sh... |

12 | Akansu, “A robust scheme for oblivious detection of watermarks/data hiding in still images
- Ramkumar, N
- 1998
(Show Context)
Citation Context ...samples or coefficients and , we may generate the marked version and so that a “1” is embedded by forcing and a “0” is embedded by forcing [36]. We can also use enforcing signs to embed one bit [37], =-=[38]-=-. Many schemes that use noncoherent detection3 belong to this Type-II category. It is the deterministically enforced relationship on that removes the need of using the host signal . For convenience, w... |

11 | Watermarking and Digital Signature Techniques for Multimedia Authentication and Copyright Protection
- LIN
- 2000
(Show Context)
Citation Context ...er researchers studied the bounds of embedding capacity under blind detection [23]–[25]. Zero-error capacity has been studied for a watermark-based authentication system under magnitude-bounded noise =-=[17]-=-, using the principles originally proposed by Shannon [26], [27]. In [28], Costa showed that the channel capacity under two additive Gaussian noises with one known to the sender equals to the capacity... |

10 |
Digital Watermarking using Shuffling
- Wu, Liu
- 1999
(Show Context)
Citation Context ...at detection. Shuffling can be considered as a permutation, which can be either random or nonrandom10 . We shall focus on the case of random permutation, where all permutations are equiprobable [18], =-=[45]-=-. We will show the effectiveness of this approach by examining the distribution of embeddable DCT coefficients before and after a random shuffling. 1) Equalizing Embedding Capacity Via Shuffling: The ... |

9 | Data hiding in image and video: Part II-Designs and applications
- Wu, Liu
- 2003
(Show Context)
Citation Context ...constant embedding rate with shuffling and using variable embedding rate with embedded control bits. We verify the effectiveness of our proposed solutions through analysis and simulation. And Part II =-=[1]-=- will apply these solutions to specific design problems for embedding data in grayscale and color images and video. Index Terms—Data hiding, digital watermarking, embedding capacity, modulation and mu... |

7 |
Resilient Video Transmission and Multimedia Database Applications
- Zeng
- 1997
(Show Context)
Citation Context ...ount of modification usually depends on both robustness and imperceptibility requirements. For example, a DCT coefficient whose magnitude is smaller than a threshold may be considered as unembeddable =-=[12]-=-. To embed as many bits as possible in each region, the number of actually embedded bits would vary significantly from region to region, and this side information has to be conveyed to the detector fo... |

4 |
Tewfik: “Self-synchronizing Watermarking Techniques
- Alghoniemy, H
- 1999
(Show Context)
Citation Context ...is reduced as increases, causing a decrease in detection reliability. For Type-II, the embedding of multiple bits can be done by enforcing relations along several mutually orthogonal directions [43], =-=[44]-=-. B. Comparison of Modulation/Multiplexing Techniques Applicability: The applicability of a particular technique depends on the type of multimedia sources and the embedding mechanism being used. Ampli... |

2 | Privacy Protection, Special Issue - Copyright - 1998 |

2 |
Blahut: Theory and Practice of Data Transmission Codes, 2nd Edition (draft
- E
- 1997
(Show Context)
Citation Context ...hen all but very few blocks having no embeddable coefficients, the bits to be embedded in the blocks with no embeddables can be treated by a detector as erasure bits. Applying error correction coding =-=[49]-=- with moderate correction capability to the data to be embedded will be able to handle this problem. Another approach to handle bad shuffle is to generate a set of candidate shuffles which are signifi... |

1 | Information Hiding: An Annotated Bibliography. [Online]. Available: http://www.cl.cam.ac.uk/ ~fapp2/ steganography/ bibliography - Anderson, Petitcolas - 1999 |

1 | Capacity of the additive steganographic channel. [Online]. Available: http://www.eecis.udel.edu/ ~marvel - Marvel, Boncelet - 1999 |

1 |
Scalar Costa scheme for information embedding. [Online]. Available: http:// wwwnt.e-technik.uni-erlangen.de/~eggers/ publications.html
- Eggers, Bauml, et al.
- 2002
(Show Context)
Citation Context ...er than the watermark. Motivated by Costa’s information theoretical result [28], distortion compensation has been proposed to be incorporated into quantization-based enforcement embedding [30], [39], =-=[40]-=-, where the enforcement is combined linearly with the host signal to form a watermarked signal. The optimal scaling factor is a function of WNR and will increase the number of bits that can be embedde... |

1 |
Optimal differential engergy watermarking of DCT encoded images and video
- Langelaar, Lagendijk
- 2001
(Show Context)
Citation Context ...erence watermark [50], [51]. Besides applying shuffling to the entire set of samples or coefficients, we can shuffle on block basis by permuting all samples/ coefficients in the same block as a whole =-=[52]-=-. We can also apply different shuffles to each frequency band of a block-based transform so that the coefficients of a particular frequency band remains in their original frequency band but permuted t... |

1 |
Data hiding in binary images,” IEEE Trans. Multimedia, 2003, to be published. Min Wu (S’95–M’01) received the B.E. degree in electrical engineering and the B.A. degree in economics from Tsinghua University
- Wu, Liu
- 1996
(Show Context)
Citation Context ...vial. For example, it is difficult to find in a binary image many overlapped but orthogonal directions to produce features that are manipulable within the just-noticeable-difference (JND) range [53], =-=[54]-=-; to obtain such directions for audio also requires a large window of samples, which could lead to significant processing delay. TDM versus CDM: TDM and orthogonal CDM are equivalent in terms of energ... |