Is blind image steganalysis practical using feature-based classification?

Steganalysis is a known practice created to detect hidden data within covered items such as images or texts. Many researches claimed that features extraction is among the suitable methods to detect hidden data within images. Mutual studies have proven that statistical tests are a good way to detect blind steganalysis. Therefore, this paper-work verify the claim proving practicality testing analysis of steganalysis system that depicts the existence of hidden data focused on gray images, by using statistical features and artificial neural network techniques. The proposed system is built to work as blind image steganalysis scheme representing common security as looked-for the most. The research basic gray level co-occurrence matrix (GLCM) displayed the properties of correlation, contrast, homogeneity, and energy in the feature set, as focused used for analyzing this study. The research experimentations adopted LSB steganography technique to create stego-images for testing the steganalysis evaluation practicality. Additionally, machine learning (ML), radial basis function (RBF), as well as the naive bayes classifiers were determined to categorize the remarks for improving detection accuracy. From the investigational results, the proposed system exemplified reliability, and enhancement in the detection rate for most steganographic methods. Further, the correlation features displayed increased accuracy with the RBF and naive bays classifiers showing steganalysis practicality in an attractive remarking contribution.