Enhancement of a Spatial Domain Data Hiding Algorithm Using Weighted Vector Filtering and Singular Spectrum Analysis

Abstract

The current state of information technology development necessitates the implementation of highly effective methods and tools to ensure the integrity and confidentiality of digital data. Today, alongside with the traditional cryptographic methods of information protection, steganographic approaches are gaining particular importance, as their main advantage lies in the ability to conceal not only the content of the data but also the very fact of its transmission. Among steganographic methods, the most widely used are techniques for embedding data in the spatial domain of raster images, which is explained by their relative ease of implementation and low computational complexity. At the same time, spatial steganographic methods are vulnerable to various distortions such as filtering, noise attacks, compression, or other image processing operations, which can reduce the robustness of hidden information and increase the risk of data loss. This highlights the need to improve existing methods to enhance their reliability and resilience against malicious impacts.

In this regard, the present work proposes a modified steganographic algorithm for information hiding in the spatial domain using Weighted Median Filtering (WMF) and Two-Dimensional Singular Spectrum Analysis (2D-SSA). This approach allows effective extraction of structural components of the image and improves the robustness of embedded data without significantly degrading the visual quality of the cover image. Furthermore, the proposed method has been extended by incorporating algorithms aimed at reducing the effects of noise (Gaussian, Salt & Pepper) and enhancing robustness against JPEG compression. To assess the quality of the embedded information, PSNR and SSIM metrics were applied, enabling quantitative comparison of the efficiency of different steganographic approaches, including methods based on Discrete Cosine Transform (DCT) and Least Significant Bit (LSB) substitution. Additionally, an adaptive watermark embedding mechanism has been implemented, which further enhances the reliability of hidden data preservation under real-world conditions.