An Adaptive Steganography Scheme Based on Visual Quality and Embedding Capacity Improvement

Abstract

In this paper, a steganography technique using LSB substitution and PVD method is presented as an adaptive scheme in the spatial domain. Our method partitions the grayscale image into several non-overlapping blocks with three consecutive pixels. The embedding algorithm can both replace the secret data with the LSBs of the middle pixel and embed it in the difference values between the middle pixel and its two neighboring pixels of the cover-block. The number of secret bits is determined adaptively based on the range divisions for embedding in the difference value. We define a new range division on gray level which takes into account a larger embedding capacity for bits. After the embedding, the proposed method detects the pixels which are sensitive to hyper distortion. Then, the embedding process will be repeated to produce insignificant visual distortion in those pixels. Our experimental results demonstrate that this iterative steganography scheme prevents significant visual distortion into stego-image. The generated PSNR values are higher than the corresponding values of the most commonly used methods, discussed in this study. Furthermore, the experimental results show that the hiding capacity increased enormously when the proposed range division is used. Finally, we illustrate that the method can pass RS and steganalysis detector attacks. DOI: http://dx.doi.org/10.11591/ijece.v4i4.6307