Blind detection resistant steganographic algorithm for images based on quincunx sampling lifting scheme

Based on analysis of the principles of blind detection techniques, a data hiding algorithm by modifying morphological wavelet high frequency coefficients is proposed. The quincunx sampling lifting scheme is used for image decomposition. Then the subband coefficients above a certain threshold are chosen for data embedding, and the embedding information table is built for amending algorithm. Moreover, the histogram adjustment strategy is introduced at the location of threshold coefficients to preserve the histogram of wavelet coefficients. Since most blind detection algorithms select classifying features according to the differences of statistical distributions between cover and stego images, the proposed method can resist the attack of blind detection techniques. Experimental results show that the proposed method outperforms previous steganographic methods, such as least significant bit (LSB) matching and pixel-value differencing in the capability of resisting current typical universal blind detecting methods.