Comparison between Neural Network Steganalysis and Linear Classification Method Stegdetect
Jiří Hološka, Zuzana Komínková Oplatková, Ivan Zelinka, Roman Šenkeřík · 2010
Steganography is an additional method leading to better securing messages up which goes hand by hand with the cryptography. This is the reason why revealing of such a message is difficult because a final steganogram uses multimedia or other transportation media along with genuine functionality. This paper deals with a blind steganalysis based on a universal neural network classification and compares it to Stegdetect - a linear classification tool. The results show that neural networks were better than the linear classification tool. The worst result was 1% in the case of neural network compared to Stegdetect where 4% was normal and 7.5% was the worst one on the same samples.