Investigation of the gain effect on the distortion degree and the efficiency of extracting a digital watermark on full-color images by the discrete transformations method

The problem of applying a digital watermark based on discrete wavelet and co-sine transformations method to a full-color image is considered. This method is a procedure for embedding a label into the frequency domain of a digital image in the algorithm of which a certain gain is used. General information about the algorithm and its components is presented. The influence of the amplifying coefficient on the effectiveness of the introduction and extraction of the watermark and the degree of distortion of the processed image is described. Studies of these dependencies have been carried out in order to find the optimal spectrum (coefficient) of the digital color space for the introduction of a hidden label into a full-color image with an optimal gain value. The Pearson linear correlation coefficient was used to measure the degree of extraction efficiency, and the metric of the index of structural similarity of digital images was used to measure the degree of distortion. The spectra that are most suitable as a digital watermark carrier were identified, as well as the optimal values of the amplifying coefficient for these spectra. The use of the optimal coefficient does not lead to significant distortion of the image, while having sufficient efficiency of watermark extraction.