A Novel Image Steganography Using Multiple LSB Substitution and Pixel Randomization Using Stern-Brocot Sequence

Abstract

Steganography is a technology of hiding confidential information within regularly used data files such as image, audio, video files. There have been several research works on the image steganography. Most of the works used only single LSB or multiple LSB of RGB components of each pixel. However, using only one LSB of the color components is an obstacle to convey much information. Again, if the pixels are not selected randomly, then the pixels used in steganography can be predicted very easily, thereby exposing the hidden information. In this paper, we have proposed a novel approach to inject information into an image in such a way that it is very difficult (almost impossible) for others (except sender and receiver) to detect and extract the hidden information within an image. We have encrypted our secret message before embedding it within an image. Pixels of the image have been selected in a randomized way using Stern-Brocot Sequence. Multiple LSBs of colors components (RGB) are used while inserting the encrypted message into the randomly selected pixels. We have implemented our proposed algorithm and showed our experimental results on the basis of PSNR and MSE for the stego images. Our results show that the quality of the images changes with different size of our secret messages and the quality of the stego image does not change significantly after applying our proposed algorithm. We have successfully extracted our secret messages from the stego images at the receiver side. Our proposed algorithm of image steganography can be useful for establishing a secret communication over a public communication using that public communication. As our proposed algorithm provides two step protection layer, it is quite impossible to extract the actual secret message from the image.