Steganalysis via a Convolutional Neural Network using Large Convolution Filters for Embedding Process with Same Stego Key

25 May 2016 Jean-François Couchot Raphaël Couturier Christophe Guyeux Michel Salomon

For the past few years, in the race between image steganography and steganalysis, deep learning has emerged as a very promising alternative to steganalyzer approaches based on rich image models combined with ensemble classifiers. A key knowledge of image steganalyzer, which combines relevant image features and innovative classification procedures, can be deduced by a deep learning approach called Convolutional Neural Networks (CNN)... (read more)

PDF Abstract