Blind image steganalysis using features extraction and machine learning
Houcemeddine Hermassi · 2021
The main theme of this work is the steganalysis of image steganography techniques by classification and feature extraction methods. Our contribution is to improve a steganalysis algorithm and implement it to extract the characteristics of the Cover and Stego images. Thus the use of a modern Machine Learning “Ensemble Classifier” is presented here as a powerful development tool that allows the rapid construction of steganographic detectors with significantly improved detection accuracy in a wide range of integration methods. The power of the proposed framework is demonstrated by a steganographic method nsF5 which masks one or more messages in JPEG images.