Towards A More Secure Least Significant Bit (LSB) Steganographic Method for Information Hiding

Abstract

Although the LSB algorithm for information hiding exploits the weakness of the human vision sensitivity (HVS) to hide secret data in a cover medium, it is relatively vulnerable to statistical steganalysis thus increasing the possibility of an eavesdropper decoding and retrieving the hidden message. This is partly aided by the sequential nature in which the algorithm replaces the target least significant bits of the cover image with the bits of the secret message making it easy for a steganalysis tool to zero in the image bits which are out of pattern with the rest. To enhance the conventional LSB algorithm imperceptibility to steganalysis attacks, an embedding procedure that utilizes the Mersenne Twister (MT) pseudo-random number generator (PRNG) to help in spreading the secret data bits over the cover image in a non-sequential manner is proposed. Using the MT PRNG makes sure that the generated sequence of numbers is repeatable, has known mathematical properties and can be implemented without needing any specialist hardware. The idea is to generate a group of random numbers of length equal to the secret message length. This series of numbers are then used to identify the target pixels color channel bits for the purpose of embedding the secret message bits. Though this approach slightly reduces the capacity of a cover image for equal payloads, it significantly increases the imperceptibility and the security of the hidden information which was the prime focus of this study.