Image Steganography by LSB Substitution and Optimal Key Permutation Using Genetic Algorithm

Steganography is an important data-hiding technique in which information is secretly passed between a sender and a receiver. In image steganography, a secret-message is embedded into a cover-image to create a stego-image. Furthermore, least significant bit substitution (LSB) is the most common technique used in image steganography. In LSB, either all or some of the pixels in the last four bits are replaced with pixels/a bit of the secret-message. Adding to its security issue, this technique is rather inefficient as it often results in a low-quality stego-image due to the high Mean Square Error (MSE) between the cover and stego-images. Consequently, two methods have been proposed to improve the stego-image’s quality: the first method involves using a Genetic Algorithm to uncover the optimal key permutation for embedding, while the second method involves elitism selection. This study adds to these by proposing an innovative method aimed at improving the stego-image’s quality. The method proposed by this study is rooted in both LSB substitution and optimal key permutation by Genetic Algorithm. First, the secret-image and the cover-image are selected, the secret-image is then converted to blocks, which are then encrypted and subsequently shuffled so as to make the embedded secret message meaningless to anyone without the encryption key. The method uses a genetic algorithm to calculate the optimal key permutation to improve the stego-image’s quality during LSB substitution. Using the method it proposes, the study found that the stego-image quality was improved substantially following the embedding of the “tiff” secret-image in the following cover-images: Baboon, Lena, Barbara, and Pepper. The study also found that the proposed method significantly reduced the computational complexity in the genetic algorithm.