Since then the papers are mainly on deriving more descriptive features and training using modern machine learning methods (e.g., ensemble classifiers). So the base "recipe" could be the papers I highlighted above.
To be fair, Hany Farid along with his student Lyu have had a prior work on such topic using higher order statistics and SVM.
Siwei Lyu. Steganalysis using color wavelet statistics and one-class support vector machines. Master’s thesis, Dartmouth College, Hanover, NH, USA, 2002.
Or in its lighter version:
Siwei Lyu and Hany Farid. Detecting hidden messages using higher-order statistics and support vector machines. In Proceedings of the Fifth Intl. Workshop on Information Hiding , pages 340–354, Noordwijkerhout, The Netherlands, 2002. S
Thank you for your propositions. I downloaded 2 papers from the first message and the ”lighter version” from the second. I will study them to do my implementation.
Thank you also for your information. The webpage with implementation seems very interesting. I have several papers with theoretical discussions but they are not friendly for people who want to write own implementation.