Active steganalysis of spread spectrum image steganography

We propose two active steganalysis schemes for spread spectrum image steganography. One is a simple estimate and subtract type algorithm that does not exploit higher order statistics while the other one is more sophisticated. We present conditions for successful steganalysis along with experimental results. Experiments show that the second method can extract up to 70% of message bits for large message sizes while the simple scheme saturates at 45%.