Multi-Domain Steganalysis Preprocessing to Fusion Feature for Optimal Stack Ensemble Model

The field of image-based steganography has been widely used, because of the advancement of steganography methods and their applications. In today's world, image-based exploits are used by the steganography approaches in the publicly available dataset. This dataset is used for data modeling to tune the model for high accuracy, robustness, and other best-fit parameters. So, this paper aims to introduce a novel way of approaching hybrid-based steganalysis, including two algorithm blocks. The first block consists of JPEG-based pre-processing as an initial-level stego cross-verification match using multi-domain steganalysis such as statistical, structural, and frequency. The second block consists of a custom-based fusion feature extraction and meta-feature analysis stage based on the statistical measure evaluation with the machine-level models and their stacked ensemble classification, which improved the analysis of the stego and cover images. As a result, our approach would be lightweight for integration modules for different areas like the initial level for data security to minimize individual and organizational hardware-level stego JPEG-image-based exploits with exception flow management, our model will enhance computational efficiency and higher performance scores of steganalysis.