A novel steganographic method using 8-neighboring PVD (8nPVD) and LSB substitution

Enhancing the Imperceptibility of Image Steganography for Information Hiding

“…Table I shows the embedding capacity (in bits) as well as the corresponding PSNR values and CWSSIM indexes using the competing approaches. Table I shows that the proposed approach outperforms the approaches shown in [10] and [13], by an average of 1.4 dB and 0.9 dB, respectively, in terms of the standard PSNR value. As well, the proposed approach surpasses the competing approaches by an average of 11.6% and 8.5% in terms of the CWSSIM index.…”

“…In addition, Table I shows that the bit-embedding capacity using the proposed approach outperforms the approach in [10] by an average increase of 13.6%. While the proposed approach is comparable to that in [13].…”

“…This section presents the evaluation metrics and the experiential results of the proposed image steganography approach compared to competing approaches shown in [10] and [13]. A set of standard gray-level images [14] with a different sizes have been chosen as cover images to check on the effectiveness of the competing approaches.…”

“…Ultimately, the stego-image is generated with a certain image quality. So, the spatial domain-based approaches can be partitioned into three categories: (i) high embedding capacity approaches with barely acceptable image quality (e.g., [5]- [7]), (ii) high image quality approaches with reasonable hiding capacity (e.g., [8]- [10]), and (iii) restricted embedding capacity approaches with a slight distortion in the image (e.g., [11]- [13]). …”

“…Encryption is performed to sensitive contents using iterative magic matrix-based encryption algorithm. In [10], an image steganographic method is presented based on the LSB substitution with a typical pixel value differencing (PVD) method, a modified version of PVD method, and an 8-neighboring (8nPVD) method, respectively, for gray-scale cover image in order to improve the embedding capacity with a reasonable imperceptible stego-image. In [13], the image is partitioned into 2×2 pixel blocks in a non-overlapping fashion and scanned in raster-scan order having correlated the leftupper and bottom-right corner pixels.…”