Image steganography using redundant discrete wavelet transform and QR factorization

Transform domain image steganography using RDWT and QR decomposition is demonstrated.Cover selection measure based on statistical texture properties is proposed. It helps to minimize detectability and improves security.QR decomposition has lower computational complexity and avoids false positive issue in SVD based methods.The redundancy in shift invariant RDWT helps to maximize embedding capacity and is robust to additive noise. The paper demonstrates image steganography using redundant discrete wavelet transform (RDWT) and QR factorization. RDWT helps to overcome artefacts due to variation in energy distribution caused by shifts in input signal. Also, when compared with SVD, QR factorization provides less computational complexity and avoids false positive issue; a major flaw in singular value decomposition (SVD) based data hiding methods. In addition, this study also proposes cover selection measure based on statistical texture analysis. Cover selection helps to enhance security of steganographic technique. Suitable cover is decomposed by RDWT and the scrambled secret image is embedded in QR factorised sub-band. Several experiments and comparative studies are performed to show efficacy of proposed method in terms of imperceptibility, capacity and robustness to various signal processing attacks. It is also examined with wavelet based and contourlet based steganalyzer. The average detection accuracy is found to be 52% and 56% respectively indicating poor ability of steganalyzer to identify stego images.