Image steganalysis method based on integrated GoogleNet

Aiming at the problem that the existing image steganalysis methods are prone to high deviation, an integrated Google net (EGN) model is constructed for image Steganalysis. The model is based on Google net, uses the mutation method to generate multiple variant networks, and uses the idea of integrated learning to integrate these networks Neural network models are integrated for image steganalysis. The research shows that the model is consistent with the existing image steganalysis methods and single Compared with Google net, it effectively improves the accuracy of image steganalysis and detection. For the image of s-uniward embedding algorithm with embedding rate of 0.4 BPP The detection accuracy of steganalysis is 96.18%.