Full band watermarking in DCT domain with Weibull model

Digital image watermarking method based on DCT and fractal encoding

“…Fazli and Moeini [34] fused DWT, SCT and SVD to construct a robust digital image watermarking method that can be applied against and correct main geometric attacks. Dong et al [35] adopted a Weibull model to analyse full band digital watermarking in DCT domain. Panah et al [36] reviewed properties of non-media digital watermarking.…”

Statistical characterisation of block variance and AC DCT coefficients for power law enhanced images

“…The JS value at location (0, 1) of AC DCT coefficients for original and power law enhanced JPEG compressed images at different qualities lies between ∼0.002 and 0.04. F = l α g , l α I (0, 1) , l α I (0, 2) , l α I (0, 3) , l α I (1, 0) , l α I (1, 1) , l α I (1, 2) , l α I (2, 0) , l α I (2, 1) , l α I (3, 0) , l α I (3, 1) , l α I (4, 0) , l α I (6, 6) , l α I (6, 7) , l α I (7,7) .…”

“…One of the important areas of this study is statistically characterising alternating current (AC) DCT coefficients of an image. The statistical characterisation of AC DCT coefficients distribution has found applications in many areas such as design of quantisers [1], image enhancement [2,3], design of image and video coders [4,5], development of robust watermarking [6,7], steganography [8,9] and camera forensics [10,11].…”

Blind Image Watermark Decoder in UDTCWT Domain Using Weibull Mixtures-Based Vector HMT

“…So far, the commonly used transforms include discrete Wavelet transform (DWT) [1]- [8], Contourlet transform [9]- [16], dual tree complex Wavelet transform (DT-CWT) [17], nonsubsampled Shearlet transform (NSST) [18], nonsubsampled Contourlet transform (NSCT) [19], discrete Cosine transform (DCT) [20], and discrete Shearlet transform (DST) [21]. The usually adopted statistical models include Gaussian distribution [5], Gaussian mixture model (GMM) [3], bivariate Gaussian distribution [15], general Gaussian distribution (GGD) [7], [16], normal inverse Gaussian distribution (NIG) [9], [12], Bessel-K form (BKF) distribution [10], [18], Gamma distributions [17], Rayleigh distributions [17], Weibull distributions [17], [20], two dimensional generalized autoregressive conditional heteroscedasticity (2D-GARCH) model [2], [13], t Location-Scale distribution [11], [14], Cauchy distributions [19], vector-based Gaussian HMT model [6], [8] and Laplacian distribution [21]. The two most common methods of digital watermark embedding are addition [2], [4], [7], [8], [11], [13], [17], [19] and multiplication [1], [6], [9],…”

“…The usually adopted statistical models include Gaussian distribution [5], Gaussian mixture model (GMM) [3], bivariate Gaussian distribution [15], general Gaussian distribution (GGD) [7], [16], normal inverse Gaussian distribution (NIG) [9], [12], Bessel-K form (BKF) distribution [10], [18], Gamma distributions [17], Rayleigh distributions [17], Weibull distributions [17], [20], two dimensional generalized autoregressive conditional heteroscedasticity (2D-GARCH) model [2], [13], t Location-Scale distribution [11], [14], Cauchy distributions [19], vector-based Gaussian HMT model [6], [8] and Laplacian distribution [21]. The two most common methods of digital watermark embedding are addition [2], [4], [7], [8], [11], [13], [17], [19] and multiplication [1], [6], [9], [10], [12], [14], [15], [16], [20], [21]. Despite the tremendous benefits, most existing transform domain multiplicative watermarking methods based on statistical modeling generally suffer from four disadvantages.…”