Digital watermarking of images using neural networks

Hwang, Min‐Shiang; Chang, Chin‐Chen; Hwang, Kuo-Feng

doi:10.1117/1.1289357

Cited by 36 publications

(14 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We establish the relationship among coefficients in the discrete wavelet decomposition of the image by using the Back-Propagation Neural Networks (BPNN) model [12]. For a selected coefficient, the network is trained with its 3×3 neighbors as input vector and the value of the coefficient as output.…”

Section: Neural Network For Video Watermarkingmentioning

confidence: 99%

Video Watermarking Based on Neural Networks

El'arbi¹,

Amar

Nicolas

2006

2006 IEEE International Conference on Multimedia and Expo

View full text Add to dashboard Cite

In this paper, we propose a novel digital video watermarking scheme based on multi resolution motion estimation and artificial neural network. A multi resolution motion estimation algorithm is adopted to preferentially allocate the watermark to coefficients containing motion. In addition, embedding and extraction of the watermark are based on the relationship between a wavelet coefficient and its neighbor's. A neural network is given to memorize the relationships between coefficients in a 3x3 block of the image. Experimental results show that embedding watermark where picture content is moving is less perceptible. Further, it shows that the proposed scheme is robust against common video processing attacks. INTRODUCTIONThe sudden increase in watermarking interest is most likely due to the increase in concern over copyright protection of content. With the rapid growth of the Internet and the multimedia systems in distributed environments, digital data owners can easy transfer their multimedia documents across the Internet. However, current technology does not protect their copyrights properly. This leads to wide interest of multimedia security and multimedia copyright protection and it has become a great concern to copyrights owners in recent years. In the early days, encryption and control access techniques were used to protect the ownership of media. Digital watermarking, the art of hiding information in a robust and invisible manner, has been investigated as a complementary technology [1]. There are some desirable characteristics of effective watermarking techniques, including imperceptibility, robustness, and security [2]. Imperceptibility is the degree of perceptual similarity between the original and watermarked signals. It is desirable to embed the watermark in a discreet, unobtrusive manner so that the watermark is imperceptible under casual observation. Robustness is the resilience of the watermark against manipulations such as lossy compression, linear and nonlinear filtering, scaling, and cropping (Stirmark [3], dewatermarking attack [4]).Security is the ability of the watermark to resist hostile attacks. Attacks are not limited to removal of the watermark, but include watermark estimation or forgery, collusion, and ambiguity attacks.Obviously, it is desirable to have an imperceptible, robust, and secure watermarking technique.Digital watermarking has recently been extended from still images to video. Many algorithms have already been proposed in the literature. It goes from the simple adaptation of watermarking algorithm initially designed for still images [5] to the definition of specific video watermarking scheme [6]. Open paths still remain in video watermarking. This technology is indeed in its infancy and is far from being as mature as for still images.In this paper, we propose a novel approach to neural network watermarking for uncompressed video in the wavelet domain. The watermark is embedded multiple times into the host data. Before actually embedding the watermark information, it take...

show abstract

Section: Neural Network For Video Watermarkingmentioning

confidence: 99%

Video Watermarking Based on Neural Networks

El'arbi¹,

Amar

Nicolas

2006

2006 IEEE International Conference on Multimedia and Expo

View full text Add to dashboard Cite

show abstract

“…Neural networks have been applied to analyze the characteristics of digital image. Back propagation neural network (BPN) model is used in Hwang et al (2000) where first nine AC DCT coefficients are the input vectors and the twelfth AC coefficients is the output vector. The weight values can be modified by the training set according to the approximate errors and binary watermark is embedded by applying the concept similar to that of pixel value difference.…”

Section: Review Of Related Work Limitations and Scope Of The Presenmentioning

confidence: 99%

Genetic algorithms for optimality of data hiding in digital images

Maity

Kundu

2008

Soft Comput

View full text Add to dashboard Cite

This paper investigates the scope of usage of Genetic Algorithms (GA) for data hiding in digital images. The tool has been explored in this topic of research to achieve an optimal solution in multidimensional nonlinear problem of conflicting nature that exists among imperceptibility, robustness, security and payload capacity. Two spatial domain data hiding methods are proposed where GA is used separately for (i) improvement in detection and (ii) optimal imperceptibility of hidden data in digital images respectively. In the first method, GA is used to achieve a set of parameter values (used as Key) to represent optimally the derived watermark in the form of approximate difference signal used for embedding. In the second method, GA is used for finding out values of parameters, namely reference amplitude (A) and modulation index (µ) both with linear and non linear transformation functions, for achieving the optimal data imperceptibility. Results on robustness for both the methods against linear, non linear filtering, noise addition, and lossy compression as well as statistical invisibility of the hidden data are reported here for some benchmark images.

show abstract

“…The number of parity check bits is 3, so the value of M is 3. The parity check bits to embed into pixel B are (100) 2 . We assume that the random number R i here is 121.…”

Section: The Embedding Proceduresmentioning

confidence: 99%

“…In reality, image watermarking techniques have come through a history of several years, and they have gone down different routes. For example, in terms of robustness, image watermarking techniques can be classified as either robust image watermarking techniques [1][2][3][4][5][6] or fragile ones [7][8][9][10][11][12].…”

Section: Introductionmentioning

confidence: 99%