Abstract:
In this paper, we introduce a muItire.solution watermarking methcd for copyright protection Of digital 1mages. The method is based un the discrete wavelet transform. A nuise type Gaussian sequence ‹s used us watermark. Tu embed the watermark robusdy and imperceptibty, watermark components are aJded tu the significant coefficients of each Selected subband by considering the human visual system (HV$) characteristics. ome smalI modifications are performed to improve the HVS model. The host image i.s needed in watermark c xtraclion procedure, and Normalized Correlation Function (NCF) is used to measure similarities of extracted watermarks. It is shown that this method is robust against wide variety of attacks. Comparison with the ex isting methods shows the better performanceof this suggested method.
Machine summary:
Due to the excellent time-frequency t'eatures and the well-ixiatching to the human visual system (HVS) characteristics, wavelet has been widely used for digital watermarking, especially after the wavelet transform became the basic method in JPEG2000 standards several years ago.
In most wavelet domain watermarking schemes, watermark is embedded into the middle-rrequency subbands coeFficientr, for two reasons: one is that low frequency components have more effects on the image quality than middle and high frequency components; the other is that high frequency components are easily removed after low pass filtering.
Authors in [ II j have proposed a wavelet based multiresolution Inethod using a human visual system, with the number ot‘ watermarks embedded proportional to the energy contained in each band.
WATERMARKING METHOD The proposed method embeds watermark by decomposing the host image using wavelet transforms.
A visual mask based on HVS characteristics is used t'or calculating the weight tactor for each wavelet coefficient of the host image and the significant coefficients from each subband are selected based on these weight factors.
The threshold T can be determined by calculating 1"alse positive probability function (Prr) according to the equation: (View the image of this page) where N is the length of the watermark [16 EXPERIMENTAL RESULTS To test the performance of’ the propo.
(View the image of this page) Figures 12 to 14 show the results of detecting the watermark with the proposed method on low-pass filtered images.
(View the image of this page) CONCLUSIONS In this paper, we introduced a inultiresoltition watermarking method using the discrete wavelet transform.