A secure image steganography technique with moderately higher significant bit embedding

Abstract

Steganography is a process to hide secret data into a cover media in an imperceptible manner. In the spatial domain of image steganography, the most common technique is Least Significant Bit Replacement (LSBR). However, LSBR is extremely sensitive to compression attacks involving truncation of LSBs. As a possible solution to the drawback of the traditional LSBR scheme, this paper proposes an image steganography technique that embeds secret data in the moderately higher significant bits such as 4th or 5th bit of a pixel. The proposed method uses a color image as a cover and according to pixel values; three groups of pixels are maintained. These groups are used for selecting the candidate pixels for 4th or 5th bit embedding. It also implements an optimal pixel adjustment process (OPAP) to minimize the visual distortion due to embedding. In addition to the OPAP, a method for randomly dispersing the secret data bits is also implemented making it harder for an adversary to detect hidden information. The experimental results for proposed method showed high values for Peak Signal to Noise Ratio (PSNR) signifying High stego-image fidelity.