There has been an increasing trend in the data exchanges over the media and an extensive use of digital media over the past decades. The pressing need in reference to digital water marking over the period has been the increasing need for individual rights protection so called copyright protection. The application for digital watermarking in the broadcast monitoring, finger printing video authentication and copy control has been on the rise. The chief aspects of information hiding constitute capacity, robustness and security. Security refers to the skill of anyone detecting the information whereas robustness refers to the resistance to the modification of the content of the cover before hidden information is finally destroyed. The Algorithms for video watermarking prefer robustness. In case of a robust algorithm, it is not possible to remove the watermark without degrading the cover content. This paper introduces the notion of video watermarking and the required features for a robust watermarked video for valuable application and focus on domains of the video watermarking techniques.
DWT, Robust Techniques, SVD, Video watermarking.
Video watermarking remains a you filed that is fast growing in the field of multimedia. There are several factors that have contributed to this growth and the list below comprises of them all:
a) The copying of digital media has become comparatively easy
b) This is a time where the need to protect intellectual rights infringements has been high
c) Malicious attacks should not erode copyright protection
d) Tempering of the digital data should be concealed at one point.
Over the last decade, video piracy has become the matter of concern for movies and studio producers. Digital video watermarking is therefore a possible means of mitigating this type of digital distribution. This is defined as the process of embedding extra information in a host video signal so that the watermark is imperceptible and with the case of a blind detection system, robust and difficult to alter. The imperceptibility of a video refers to the capability of the video in retaining its original quality after being embedded in in another video. Just like in most applications, the original video is not present during the extraction of the watermark. We say that a watermark is robust if it can withstand the most common video distortions that include geometric and signal attacks.
There are several approaches that have been developed so as to curb the problems that are associated with watermarking. Some of these approaches operate in singular value decomposition domain (SVD). This is applied as the host image to find the singular values. These singular values are used to modify through adding the watermark and then the SVD is performed a new on the resultant matrix so as to calculate the values already modified. The final singular values are then replaced using the modified values so as obtain a watermarked image. An operation that is inverse is the performed at the decoder so as to extract the watermark. A watermark than combines two approaches, discrete wavelet transform (DWT) and SVD introduced. This approach only applies SVD equally to only the HL and LH sub-bands of a I-level Haar DWT other than using the entire image. The watermark is then embedded in two parts that are then embedded in singular values of the LH and HL. These schemes are impractical since the values generated from the image are required at the decoder so as to extract the image.
Loo et al. first familiarized with DT and CWT based blind schemes. Since DT and CWT potentially contain approximate shift invariance and good directional selectivity properties, they considered these schemes . This approach is, a result, becoming a popular choice. This paper proposes DT CWT and SVD based hybrid watermarking schemes which uses U channel in a YUV representation of a video frame. The changes in a U channel of a video frame are not noticeable to the human eye than those in the Y channel. Hence, the imperceptibility of the water mark is reachable at higher watermark strengths. Because of the estimated shift invariance physiognomies of the DT and CWT robustness to geometric attacks, rotation cropping is enhanced. The SVD technique is used for this particular approach because of its singular values. In a much more sense, when a small perturbation is added to a frame, the singular values change only a small bit. The watermark at the decoder is therefore removed in a blind fashion without necessarily using the initial video sequence or a remarkable as it is in the case of conventional SVD-based approaches. The experimental results obtained are prove of performance as it regards to rotation, cropping, up scaling, downscaling in revolution, noise addition and compression and aspect ratio.
Proposed water marking method
This paper proposes a method in which the watermark is embedded in the singular values of the low frequency DT CWT coefficients of the U channel. To begin with, since the watermark in the low frequency coefficients is robust against compression and geometric distortions, the three level 3 complex coefficients (Xu 3,d) are selected after performing a 3-level DT CWT decomposition on the U channel of a video frame. Worth noting is the fact that DT CWT produces six directional sub bands (d = 1, 2, • • • , 6) of complex coefficients with each level at angle of ±15° , ±45° and ±75°. This paper however exploit only four sub-bands in the direction ±15° and ±75°(d=1,2,5,6) and therefore apply the SVD to them. If the dimension of (Xu 3,d) in a direction shown is M*M then it is possible to define SVD as
Where d=1, 2, 5, 6 represent four dimensional sub-bands and Uu 3,d
and Vu3,d are the unitary matrixes while Su 3,d is a diagonal expressed in the following format.
The diagonal elements in their descending order represent the singular values of Xu 3,d. The watermark represented by ω of a frame is 2d array of elements that is pseudo-random of 1’s and -1’s that are generated using the same watermark key (K) for β consecutive frames. This key not repeated in the remaining of the video sequence. The ideal length of β is a trade-off between robustness to temporal frame averaging (TFA) and watermark remodulation (WER) threats. All the dimensions of ω need to be four times smaller than those of U channel. Because of the redundancy of the DT CWT, some of the components of the pseudo-random watermark that are found in the null space of the inverse DT CWT are possibly lost during the process of inversion. In that case, we embed the coefficients (Ww 1,d) of a 1-level DT CWT of the watermark instead of embedding it directly. The singular values of Ww 1,d for d=1,2,5,6 can be represented as below:
What exists as the major differences between the singular values in directions d = 1 and 6 and d = 2 and 5 are very minor . Consequently, we use four different sets and cluster them into pairs and lastly, so as to obtain a mean difference between the two directions that exist within two directions in a pair, the singular values of the level 3 coefficients of the U Channel require modification as follows:
Here, β is the embedding bit (either a +1 or -1) that is generated pseudo-randomly and a α is the scaling factor that acts to control the strength of the embedding watermarks. The level 3 coefficients (X ˆu 3,d) are obtained by replacing the diagonal array (Su 3,d) of (1) by S ˆu 3,d as expressed in the equation below :
Finally, through applying the inverse DT CWT on the coefficients of the watermark, a watermarked frame is obtained and the process is repeated for each frame of a video sequence.
Even though the watermark is embedded in a singular values of the 3 coefficients of a 3-level DT CWT decomposition is extracted from either of the three levels depending on the downscaled video resolution . The scheme has been adopted so as to relieve the effect of the truncation of the high frequency bands from a frame of a video especially when downscaling has occurred.
In the watermark extraction process, the SVD is applied on the embedding sub-bands (1, 2, 5 and 6) of the specific level from which the watermark is extracted as expressed in the below expression:
L= 3 or 1.2 and is the level of DT CWT decomposition and (.)’ represent the array of matrices that are approximated from the frame that is watermarked and may be subjected to attacks. On the side of the encoder, (with reference to b, whereas the singular values of the watermark are added to a sub-band of a pair, they were subtracted from the other sub-band of that pair and vice versa. Due to this, since the small mean difference between the singular values of a pair, a large difference between them was created. It therefore means that the difference between the pair is represented as below:
It should be noted that b = 0 indicates an error bit as +1 is embedded that is extracted as -1 or the vice versa. This process is repeated over a number of frames and then a cross-correlation (NCC) is performed between the patterns (b(f) and (b(f)’ in which the f stands for the frame index. Through comparing the correlation peak with the threshold. A conclusion is therefore made as to whether the video is to be watermarked.
Experimental Results in Table form
This paper proposed a blind digital video watermark approach using the U channel through the combination of the benefits of DT CWT and SVD methods. The watermark was embedded in the singular values of the level 3 constants of a 3-level DT CWT decomposition of the U channel. The realization was that the U channel allows a stronger watermark to be embedded than it is with its alternative Y channel as it maintains unmatched quality when compared with the original video.
F. Hartung and M. Kutter, “Multimedia watermarking techniques,”
Proceedings of the IEEE, vol. 87, no. 7, pp. 1079–1107, Jul. 1999.
R. Liu and T. Tan, “An SVD-based watermarking scheme for protecting rightful ownership,” IEEE Transactions on Multimedia, vol. 4, no. 1, pp. 121–128, Mar. 2002.
N. M. Makbol and B. E. Khoo, “A new robust and secure digital image watermarking scheme based on the integer wavelet transform and singular value decomposition,” Digital Signal Processing, vol. 33, no. 0, pp. 134–147, Oct. 2014.
C. C. Lai and C. C. Tsai, “Digital image watermarking using discrete wavelet transform and singular value decomposition,” IEEE Transactions on Instrumentation and Measurement, vol. 59, no. 11, pp. 3060–3063, Nov. 2010.
P. Loo and N. Kingsbury, “Digital watermarking using complex wavelets,” International Conference on Image Processing, vol. 3, pp. 29–32, Sep. 2000.
N. Kingsbury, “Shift invariant properties of the dual-tree complex wavelet transform,” International Conference on Acoustics, Speech, and Signal Processing, vol. 3, pp. 1221–1224, Mar. 1999.
M. Asikuzzaman, M. J. Alam, A. J. Lambert, and M. R. Pickering, “Imperceptible and robust blind video watermarking using chrominance embedding: A set of approaches in the DT CWT domain,” IEEE Transactions on Information Forensics and Security, vol. 9, no. 9, pp.1502–1517, Sep. 2014.
M. Asikuzzaman, M. J. Alam, A. J. Lambert, and M. R. Pickering, “A blind digital video watermarking scheme with enhanced robustness to geometric distortion,” International Conference on Digital Image Computing Techniques and Applications, pp. 1–8, Dec. 2012.
Video watermarking is the process of adding a digital watermark or logo to a video, typically as a form of protection against copyright infringement. The watermark is usually a transparent overlay that contains a symbol, text, or logo that identifies the owner or creator of the video. The watermark is embedded in the video stream and is visible throughout the video playback. Video watermarking helps to discourage unauthorized distribution or use of the video and makes it easier to track the source of any unauthorized copies that may circulate.