NEW SECURE QIM-BASED INFORMATION HIDING ALGORITHMS

Abstract

The paper proposes two information hiding algorithms for multimedia based on Quantization Index Modulation (QIM): IM-QIM and SIM-QIM. They are designed to prevent a statistical attack which is able to restore a secret key using the correlation between key bits and multimedia data samples. To obtain the required security, we use a correlation immune embedding function, which guarantees statistical independence between the watermarked data and the key. The proposed algorithms are described for the case of spatial embedding in images but the algorithms can also be used to hide information in any multimedia data source in spatio-temporal and spectral domains. The results of the experimental investigation have confirmed that the algorithms developed provide the required security against a statistical attack and also have shown that the algorithms do not introduce additional distortions compared with the conventional QIM method. However, the experiments have also shown that the new algorithms are less robust against additive noise and JPEG compression.