A Modified Least Significant Bit Randomized Embedding Method based on Image Partitioning and Columnar Transposition with Encryption
Fredilyn B. Calanda, Ariel M. Sison, Mark Rennel D. Molato, Ruji P. Medina · 2019
While traditional embedding methods which are usually based on LSBs provide security to hidden messages in image-based Steganography, steganalyses have shown these methods to be vulnerable to detection and extraction, thus compromising the security of embedded messages. In this study, a new embedding method based on LSB is introduced to address this vulnerability. By using a one-time pad encryption method, the security of secret messages before embedding is ensured. A partitioning method based on columnar transposition and RGB color plane scattering technique is employed to improve scattering and ensure randomization during embedding. The experimental result shows an excellent outcome of the Peak-signal-to-noise Ratio (PSNR), Mean square Error (MSE), and Structural Similarity (SSIM) index. The PSNR value of the 5 test images results ranges from 53.2122 to 60.8716 all are higher than the acceptable value of PSNR, which is 30db.The average MSE result is0.2766 for all color planes, which also provide an outstanding result since the lower the MSE means the accumulative squared error between the stego image and the original one is lower. The average SSIM obtained in the testing is 0.9792for the 5 test images, which is very close to 1. The result supports the claim that the resulting stego images have low distortion which means that the cover image is almost the same with the stego image and that the modified LSB embedding algorithm provides excellent image quality with high imperceptibility and provides added security layers.