The relevance. One of the areas of digital image processing, including images of Earth remote sensing, is associated with the embedding of additional information into such images for various purposes. Embedding of additional information into a digital image leads to distortion of the digital image natural model, as well as to the possible occurrence of visual artifacts. In the case of images of Earth remote sensing, such distortions may entail, for example, distortion of the object boundaries, as a result of which further analysis of the images will lead to incorrect results. Therefore, the studies aimed at finding new ways to reduce distortions caused by embedding additional information are relevant. The aim of the research is to improve the quality of steganographic embedding of information in the coefficients of discrete cosine trans form of Earth remote sensing images by developing an improved embedding algorithm based on the QIM method and ensuring the correction of distortions of the natural image model in the frequency domain. Objects: algorithms for steganographic embedding of information in the coefficients of digital images discrete cosine transform. Methods: QIM steganographic method, methods of mathematical statistics, computational experiments. Results. The paper proposes a new approach to minimizing the distortion of the natural model of a digital image in the discrete cosine transform domain, based on changing the quantization step depending on the local image characteristics in the frequency domain. The result of the work is an improved algorithm for embedding the information in the coefficients of discrete cosine transform of images of Earth remote sensing, combining this approach with an approach to ensure the correct extraction of embedded information proposed by the authors earlier. The results of the experiments show that the developed algorithm, along with ensuring the accuracy of extraction, can significantly reduce the distortions introduced into frequency coefficients and ensure statistical indistinguishability of the cover images and stego images in 75 % of cases.
