Adaptive Steganography Using 2D Gabor Filters and Ensemble Classifiers

In order to preserve the statistical properties of image in all scales and orientations when the embedding changes are constrained to the complicated texture regions, a steganography method is proposed based on 2 dimensional (2D) Gabor filters and Ensemble classifiers. First, we use histogram sequence features of filtered images generated by Gabor filters to describe the texture properties of the image. Then, we design the distortion function using the classification results differences of classifiers between cover and stego. The stego images are used to train the ensemble classifiers using the current popular steganography schemes, such HUGO and S-UNIWARD. Then, the message is embedded using STC (Syndrome Trellis Code). The experimental results show that the proposed scheme can achieve a competitive performance compared with the other steganography schemes HUGO, MVG and S-UNIWARD when using the SRM detection.