Abstract: The article contains the studies, which are aimed at improving the precision of steganalysis in the sphere of digital image DWT. The authors analyze the causes of inaccuracy of the modern stegoanalysis methods based upon the support vectors, then they offer the directions for improving the teaching quality. In order to improve the quality of teaching support vectors machine the authors study the interrelation between the spheres of one-dimensional and two-dimensional DWT and the influence of the changes in the coefficients of the high frequency spheres of the two-dimensional DWT upon the coefficient spheres of the one-dimensional DWT.  The steganographic influence involves the change in the value of the lower meaning bit coefficients of the DWT. Considering the study results the authors develop an original method, guaranteeing greater precision in the sphere of finding incorporated information in the high frequency areas of the two-dimensional DWT image.  In order to prove the precision of the original method, the authors compare it with some modern steganalysis methods. Experimental results of a comparative study prove that the original method provides for greater precision (generally 10-15% higher than other evaluated methods) when detecting the fact of steganographic influence in high frequency areas of HL and LH of the two-dimensional DWT. The original method also provides for the same precision in the high frequency HH area, as do other modern methods evaluated in this article.

Abstract: The article deals with the stability of the digital watermark built-into the image-container through the use of steganography algorithms based on discrete wavelet transform (DWT), to external influences, such as JPEG lossy compression, filtering, noise and scaling. The author states that steganographic algorithms performing embedding can provide good secrecy of the digital watermarking, and tend to use the coefficients of correlation property between different planes of one subband having the same coordinates. It is noted that an important parameter when using steganographic algorithms based on the DWT is the choice of the level of wavelet decomposition. The authors describe the methodology which was used to assess the sustainability of the digital watermarking to to external influences on the image-container and experimental conditions. The authors also evaluated stability of DWT ti JPEF lossy compression, to Gaussian white noise, to scale the image to the image filtering. In conclusion the authors say that the results of DWT stability to external influences on the image-container confirm the theoretical advantage of using low-frequency plane wavelet decomposition in steganographic systems with high demands on DWT sustainability.